Skip to main content

Full text of "Sol-gel derived silica optics"

See other formats


SOL-GEL DERIVED SILICA OPTICS 



By 
SHI-HO WANG 



A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL 

OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT 

OF THE REQUIREMENTS FOR THE DEGREE OF 

DOCTOR OF PHILOSOPHY 

UNIVERSITY OF FLORIDA 

1988 



ACKNOWLEDGMENTS 

I am deeply honored to acknowledge several persons who have helped me during the 
time of my research as a graduate student at the University of Florida and as a scientist 
at GelTech Inc., Alachua, Florida. 

I am grateful to my advisor Professor Larry L. Hench who has shared my dream of 
creating a new method for manufacturing high-tech silica optical monoliths, including 
high power glass lasers for nuclear fusion which might contribute to freeing mankind 
from energy and pollution crisises. This dream has been partially realized by this 
research and I greatly appreciate his guidance and support. 

Dennis A. LeSage, Candace E. Campbell, and Grib Murphy of GelTech Inc., and Dr. 
Jon West, Guy LaTorre, and Martin Wilson of the Advanced Materials Research Center of 
University of Florida assisted me directly or indirectly in this work. I give each of them, 
my friends, sincere thanks. My appreciation is also extended to Linton E. Floyd, III, and 
the Glass Fab Inc. for arranging and performing the gel-silica optical property proving 
tests, and to Professor Stephen F. Jacobs in Optical Sciences Center of the University of 
Arizona for the low temperature gel-silica thermal expansion test. 

Financial support from the U.S Air Force Office of Scientific Research through 
contract no. F49620-83-0072, GelTech Inc. and the Department of Materials Science 
and Engineering were very important to me and made the research and this manuscript 
possible. I am grateful to Dr. Donald R. Ulrich of the AFOSR for his understanding and 
contributions to my success. 

Special thanks are given to Professor Gholamreza J. Abbaschian, Chairman of the 
Department of Materials Science and Engineering, and Professor John Staudhammer of 
the Department of Electrical Engineering for their unforgettable assistance and 
encouragement at a very critical moment in September 1987. 

i greatly appreciate the members of my supervisory committee, Professors 
Vellayan Ramaswamy of the Department of Electrical Engineering, Joseph H. Simmons, 



David E. Clark and Gholamreza J. Abbaschian of the Department of Materials Science and 
Engineering for their advice and recommendations regarding this dissertation. The 
responsibility for any remaining errors or shortcomings is, of course, mine. 

Words are insufficient to express gratitude to my parents for their constant 
support and to my brothers and sisters for their consideration in Taiwan. I am also 
particularly indebted to my wife, Sue-Ling, not only for her great backup but also for 
her scientific discussions, and to my daughter, Jean, for understanding why we couldn't 
have much fun together while this work was being finished. 



iii 



TABLE OF CONTENTS 

Page 

ACKNOWLEDGMENTS ii 

ABSTRACT—- • ■•••• vi 

CHAPTERS 

1 INTRODUCTION TO SOL-GEL DERIVED SILICA GLASS TECHNOLOGY 1 

2 SOL-GEL TRANSFORMATION AND EXPERIMENTAL PROCEDURES 1 2 

Introduction 12 

Literature Review of Sol-Gel Transformation Modeling 1 2 

Experimental Procedure 48 

Results 57 

Conclusions 64 

3 PHYSICAL PROPERTIES OF PARTIALLY DENSIFIEDSIUCAXEROGEL 67 

Introduction 67 

Review of the Literature 68 

Experimental Procedure 73 

Results and Discussions 81 

Conclusions 1 26 

4 DEHYDRATION OF SOL-GEL DERIVED SILICA OPTICS 128 

Introduction 128 

Review of the Literature Regarding Dehydration 130 

Experimental Procedure 1 39 

Results and Discussions 145 

Conclusions 157 

5 OPTICAL PROPERTIES OF FULLY DEHYDRATED SIUCA GEL GLASS 1 6 

Introduction 1 60 

Literature Review Regarding Optical Properties of Silica Glass 161 

Experimental Procedure 1 78 

Results and Discussions 1 85 

Conclusions 203 

6 SILIGA GEL OPTICAL FILTERS USING TRANSITION-METAL COMPOUNDS ■ 206 

Introduction 206 

Review of the Literature 207 



IV 



Experimental Procedure 229 

Results and Discussions 230 

Conclusions 238 

7 CONCLUSIONS AND RECOMMENDATIONS 239 

REFERENCES 244 

BIOGRAPHICAL SKETCH 252 



Abstract of Dissertation Presented to the Graduate School 

of the University of Florida in Partial Fulfillment of the 

Requirements for the Degree of Doctor of Philosophy 



SOL-GEL DERIVED SILICA OPTICS 

By 
SHI-HO WANG 

April 1988 

Chairman: Dr. Larry L. Hench 

Major Department: Materials Science and Engineering 

Large monolithic xerogel silica glasses were successfully made from tetramethyl- 
orthosilicate and distilled water using the combination of an acidic drying control 
chemical additive (DCCA) and a specially designed drying chamber. The acidic DCCA 
increases the gel strength by formation of a fibrillar ultrastucture, and the drying 
chamber reduces the catastrophic capillary forces inside the wet gel body. 

Partially densified monolithic gels up to 850°C were routinely made for physical 
property tests and compared to commercial fused silicas. Although the mechanical 
properties of the porous gel-silica monoliths such as microhardness, Young's modulus, 
toughness, flexural strength, density are relatively lower than fused silica, the optically 
transparent porous gel silica has a uv cut-off ranging from 250-300 nm. Such a porous 
gel with excellent optical transmission and a highly uniform pore radius of 10-50 A 
offers a unique, chemically stable matrix for impregnation with a second phase of 
optically active organic or inorganic compounds. 

The processing and properties of Types I and II fused quartz optics and Types III and 
IV synthetic fused silica optics are compared with the new organometallic sol-gel 

vi 



derived gel-silica optics. Fully dehydrated and densified gel-silica has excellent 
transmission from 165 nm to 4400 nm with no OH absorption peaks. This optical 
transmission is equivalent to the best type IV fused silica. The other physical properties 
and structural characteristics of the dehydrated dense gel-silica are similar to fused 
quartz and fused silica. However, the dense gel-silica has a lower coefficient of thermal 
expansion of 2.0 x10" 7 cm/cm compared with 5.5 x 10* 7 cm/cm for standard vitreous 
silicas. The CTE value is temperature independent from 80 K to 500 K. Sol-gel silica 
optics can be made as complex shapes by casting of the sol into inexpensive plastic molds. 
Monolithic silica gel optical filters were produced by chemical doping with various 
transition-metal ions (e.g., cobalt, copper, nickel). Color changes occurred with 
various sintering temperature indicating a unique method to control light wavelength 
filtration in the visible range. For instance, the observed color or spectral (major peak 
of absorption) shifts for the 160°C, the 850°C, and the 900°C Co" ion doped gels were 
reddish pink (505 nm), deep blue (660 nm), and greenish black (670 nm) 
respectively. The optical absorption spectra of the chemically doped-silica are 
interpreted in terms of ligand-field and molecular orbital theories. 



vii 



CHAPTER 1 

INTRODUCTION TO SOL-GEL DERIVED SILICA GLASS TECHNOLOGY 

One of the world's most pervasive chemical compounds is silicon dioxide (Si02). 
This compound can exist in many forms — crystalline or amorphous, hydroxylated or 
dehydroxylated — but is most often called "silica" as a generic name. 

Silica-based compounds have been fabricated and utilized by mankind for tens of 
thousands of years, although only in the past few decades have significant strides been 
made in understanding the variables that control silica chemistry [1-7]. 

Application of this knowledge has produced many useful materials worth billions of 
dollars per year; however, today's rapidly accelerating technology demands even greater 
performance of silicate materials as well as the need to create new materials. The 
objective of this study is to produce a number of new materials using sol-gel silica 
processing, including (1) ultraporous gel monoliths for optical and chemical matrices, 
(2) ultrapure monolithic gel-glasses with ultralow optical absorption, and (3) 
chemically doped gel glass monoliths for optical filters with low expansion coefficients 
and high softening points. 

Traditional silica glasses are manufactured by melting natural quartz minerals or 
synthetic silica, or by flame or plasma vapor-deposition methods. Generally, four types 
of commercial vitreous silica are recognized and identified: Type I is obtained by electric 
melting of natural quartz in vacuum. Type II is made by flame fusion of quartz. Type III 
is made by vapor-phase hydrolysis of pure silicon tetrachloride carried out in a flame. 
Type IV is made by oxidation of pure silicon tetrachloride which is subsequently fused 
electrically or by means of a plasma. Types I and II are called fused quartz, whereas 
Types III and IV are called synthetic fused silica. 



Fused quartz is melted at temperatures above its liquidus (1713°C) from crushed 
natural crystalline quartz powders of mixed particle size, well above micrometers in 
diameter [8]. The initial size of these particles, millions of times larger than a silica 
molecule, limits the control over the resulting structure and in part determines the 
temperature necessary for melting, homogenization, and fabrication. Glass products 
from this method have numerous deficiencies; impurities, inhomogeneities, seeds and 
bubbles, a high energy requirement for raw material crushing, melting and 
homogenization, as well as premature phase separation and crystallization. 

Chemical reactions used to produce synthetic fused silica by flame hydrolysis of 
silica tetrachloride (type III) and by vacuum plasma oxidation of silica tetrachloride 
(type IV) are shown in equations #1 and #2: 

Type III (hydrolysis) 

SiCU + O2 + 2 H 2 — -> SiC-2 + 4 HCI ( 1 ) 

Type IV (oxidation) 

SiCU + O2 — -> Si0 2 + 2CI 2 (2) 

In fact, it is very difficult to have a complete reaction for either of these two 
equations. Consequently, water contents of several thousand ppm are present in type III 
silicas, and SiCl4 in few hundred ppm is retained as an unreacted residual in both type 
III and IV silicas. In addition to these two intrinsic impurities, the resultant glasses 
from type III and IV processes have extrinsic impurities in the range of few parts per 
million (ppm) due to the contamination of raw materials and crucibles at high 
temperatures (about 1900°C). 

Table 1-1 [9] lists the dominant characteristics of commercial brands of silica 
corresponding to these four types. Their transmission curves are summarized in Figure 
1-1 and Table 1-2 [10]. Type I and II glasses have more impurities (Table 1-1) which 
make uv transmission curves cut off at higher wavelengths (curves 2 and 3 in Fig. 1-1) 
than that of type III and IV glasses (curve 1 in Fig. 1-1). The amount of water (Table 1- 



Table 1-1 

Preparation and characteristics of four types of vitreous silica 



Type 


I 


II 


II! 


IV 


Process 


Electromelted 


Flame-fused 


Hydrolyzed 


Oxidized 




Quartz 


Quartz 


SiCI 4 


SiCI 4 


Example 


IR-Vitreosiia 


Herasil b 


7940° 


Spectrosil WFa 




lnfrasil b 


Homosil b 


Dynasil d 

Spectrosila 

Suprasil b 


7943c 
Suprasil-Wb 


Impurity 


(ppm): 








OH 


<5 


400-1500 


-1000 


~0(<0.4) 


Al 


30-100 


<1 


<0.2 


<0.2 


Sb 


<0.3 


<0.1 


<0.1 


<0.1 


As 








<0.02 


<0.02 


B 


4 


3 


0.1 


0.1 


Ca 


16 


0.4 


<0.1 


<0.1 


CI 








100 


up to 200 


Cr 


0.1 





0.03 


0.03 


Co 








0.0001 


0.0001 


Cu 


1 


1 


<1 


<1 


Ga 








<0.02 


<0.02 


Au 








<0.1 


<0.1 


Fe 


7 


1.5 


<0.2 


<0.2 


Li 


7 


1 








Mg 


4 











Mn 


1 


0.2 


<0.02 


<0.02 


Hg 








<0.1 


<0.1 


P 


0.01 


0.005 


<0.001 


<0.001 


K 


6 


<1 


0.1 


0.1 


Na 


9 


5 


<0.1 


<0.1 


Ti 


3 


2 








U 





0.0006 








Zn 








<0.1 


<0.1 


Zr 


3 












a: Thermal Syndicate, England. b: Heraus Amersi!, Heraeus, Sayreville, NJ. 

c: Corning Glass Work, Corning, NY. d: Dynasil; Berlin, NJ. 




240 



320 400 1000 

Wavelength nm 



3000 



5000 



Figure 1-1 Transmission curves for commercial vitreous silica 

10mm thick 



Table 1-2 
identification of transmission curves of silica glasses 



Manufacturer 


Product name 


Type 


UV curve 


IR 


curve 








in Fig. 1-1 


in 


Fig. 1-1 


Amersil, Inc. 


Herasil 


II 


3 




B 


(Heraeus) 


Infrasil 


I 


2 




C 




Homosil 


II 


3 




B 




Suprasil 
Suprasil-W 


III 
IV 


1 

1 




A 
C 


Corning Glass 
Works 


Code 7940 
Code 7943 


III 
IV 


1 
3 




A 
C 


Dynasil Corp. of 
America 


Dynasil-1000 


III 


1 




A 


Thermal Syndicate 
Ltd. 


Spectrosil 
Spectrosil WF 
IR-vitreosil 


IN 
IV 

I 


1 
1 
3 




A 
C 
C 



1) in silica glass depends on which type of process is used. For example, Spectrosil WF 
(type IV) in Table 1-1 has a water content less than 0.4 ppm compare to 1200 ppm for 
Suprasil (type III). It is observed that there is no significant shift in the uv cut off 
between type III and IV silicas due to the increased water content if the other impurities 
are constant. However, in the infrared range, water (type III, curve A in Fig. 1-1) 
noticeably gives a strong absorption. These shortcomings can limit the use of glass 
products made by the traditional techniques described above, as summarized in Table 1- 
3. 

Sol-gel processing has been used for many years, although the principal chemical 
and physical mechanisms are still not clearly understood [11-14]. In recent years 
special applications require silica optical components that meet very stringent 
requirements. The sol-gel method offers new hope in that structural manipulation is 
possible on an extremely fine scale, within the nanometer range, thereby allowing 
production of a new generation of silica materials. The outstanding features of these 
silicas include very high homogeneity, very high purity, potentially extremely low 
optical loss, ease of chemical doping, and near net shape casting. These features make 
sol-gel silicas potentially applicable to a wide range of optical products including lenses, 
mirrors, wavequides, optical fibers, integrated optoelectronics, and host materials for 
filters, lasers, and non-linear optical elements or compounds. 

The sol-gel process as it relates to silicas is summarized briefly. A sol is defined 
as a dispersion of colloids in a solvent. Silica colloids are solid particles with diameters 
ranging from 1nm to 100 nm which depend upon the type and amount of drying control 
chemical additive (DCCA) in the solution [15-19]. In this study all colloidal particles 
are synthesized by the hydrolysis of tetramethylorthosilicate (TMOS) [Si(OCH3)4] 
followed by the growth of the hydrolyzed species [Si(OCH3)4. n (OH) n with 0<n<4] [14, 
20]. 



Table 1-3 
Limits for the four types of silica fabrication processes 

Type Fabrication Limits 

Type I & II ( 1 ) Bad homogeneity (granular microstructure and bubbles) 

(2) Noticeable water content -- few tens to hundreds ppm. 

(3) High impurities -- in the range of few ppm from nature 
quartz mineral. 

(4) Micrometer scale structural manipulation -- quartz is 
ground to few micrometers before sintering. 

(5) High sintering temperature (above 1700°C) -- 

(a) High energy cost; 

(b) React with crucible, thus impurities; 

(c) Possible initiate crystallization. 

Type III (1) High water content(above 1000 ppm). 

(2) High sintering temperature (above 2000°C) -- 

(a) High energy cost; 

(b) React with crucible, thus impurities; 

(c) Possible initiate crystallization. 

Type IV ( 1 ) Detectable water content (around 1 ppm). 

(2) High sintering temperature (above 2000°C) -- 

(a) High energy cost; 

(b) React with crucible, thus impurities; 

(c) Possible initiate crystallization. 



s 



This over-saturated sol is never chemically stable in the presence of the DCCA 
and/or under thermally activated conditions; however, after some time and with the 
addition of thermal energy a sufficient concentration of colloids that are within an 
appropriate size distribution is reached and a zero surface charge is obtained [21]. At 
this point the colloids become randomly linked together in fibrillar chains with 
thermally activated Brownian motion in the presence of a Van der Waals attractive force 
and a base catalyst [see p. 224 in ref. 4]. As the chains grow they form three- 
dimensional irregular structures throughout the liquid phase. A network develops with 
the liquid phase localized within the solid gel skeleton and microscopically confined by 
it. The "sol" has lost its freedom of movement and now becomes a "gel"; this is described 
as the gelation point. 

Solids tend to decrease their interfacial area so as to minimize surface energy. 
Therefore, after the gelation point has been reached the weakly connected spherical- 
particle chains tend to minimize surface energy by particle rearrangement, thereby 
forming a strong fibrillar-shaped ultrastructure. This phenomenon continues during the 
aging process (also termed syneresis), in which liquid is expelled from the gel body and 
the weak gel shrinks and becomes stronger. 

In this study the first goal, described in Chapter 2, is the production of silica- 
based monolithic dried xerogels composed of (a) pure silica and (b) doped with 
transition-metal elements. A xerogel is defined as a gel from which the liquid phase has 
been evacuated under ambient pressures. The net size and porosity of a xerogel is 
minimized, at least to some degree, by surface energy as the liquid is removed. However, 
without the help of the DCCA in controlling the colloidal particle size, this can not be 
realized because of cracking during drying. 

In the amorphous form of silica, a tetrahedral arrangement is primarily favored 
by the radius ratios of the silicon to the oxygen ions and by the bonding of sp 3 hybrid 
orbitals in Si02- X-ray diffraction studies by Mozzi, Warren and Uhlmann [22, 23] 



9 



have shown that silicon forms bonds with oxygen of variable bond angles that are 10% 
within the 144° maximum in the distribution of Si-O-Si angles. Various arrangements 
of these SK>2 ietrahedra are possible in noncrystalline silica gels. Bonding oxygens at the 
corners of two silica ietrahedra can be easily disconnected in the presence of uneven 
hydrostatic stresses and water [24]. DCCA's can be used to minimize the particle size 
within the polymerized chain, thereby improving the strength of the gel structure so 
that during the critical drying process the gel can endure differential evaporation 
without initiating cracking. 

The processing and physical properties of dried monolithic silica xerogels, heated 
from 15CFC to 900°C, are discussed in Chapter 3. This ultraporous material has 
densities ranging from 0.7 g/cm3 to 2.10 g/cm3 depending on the initial conditions of 
the sol, such as the variation of DCCA and/or the amount of water used, as well as the 
aging and drying temperatures. 

Two types of water exist within the dried xerogel structure — chemical water and 
physical water [25], which must be removed to achieve monolithic optical components. 
Water in solution can hydrolyze the silicon-oxygen-silicon bond. The hydroxyl ion's 
oxygen is covalently bonded to silicon, whereas the hydrogen ion forms an ionic bond to 
the oxygen. Consequently, chemical water results with hydroxyl groups strongly 
attached to the gel's surface. The physical water associated with hydrogen-bonding of 
surface hydroxy! groups exists within the ultraporous space of the gel body. 

A major problem with monolithic silica xerogels, especially for high- 
transmittance optical components, is the removal of chemically bonded water, also called 
a silano! group. The chemically bonded silanols give rise to the fundamental vibration of 
hydroxyl ions occurring at a wavelength of 2669.4 nm, Also present are vibrational 
overtones and combinations of this ion and associated water occurring at the following 
wavelengths: 2919.7 nm, 2768.9 nm, 2698.3 nm, 2262.5 nm, 2207.5 nm, 1890.4 
nm, 1459.9 nm, 1408.5 nm, 1366.1 nm, 1237.9 nm, 1131.2 nm, 939.0 nm, 704.2 



10 

nm. These IR absorptions are the result of electromagnetic vibrational interactions with 
the electrons, atoms, and molecules of the gel water. Selectively absorbed light energy, 
such as this, is mostly converted into heat. Consequently it is important to reduce the 
hydroxyl groups to nondetectable levels in order to minimize absorption loss, especially 
for optical lenses, optoelectronic signal processors, optical fiber, filters, and laser 
resonant host systems. Therefore, monitoring the IR absorption peaks is a primary 
method for determining the degree of dehydration achieved during densification [26, 
27]. 

Consequently, the second goal of this study is to dehydrate and densify monolithic 
silica xerogels; this is described in Chapter 4. Two methods are investigated: (1) 
sintering samples in an air atmosphere and (2) chemical treatment and sintering in a 
controlled gas atmosphere (e.g., carbon tetrachloride). At sufficient temperatures these 
gases can react with the hydroxyl groups to form hydrogen chloride which escapes freely 
from the unclosed ultrapores [28]. The dehydrated xerogel samples are then exposed to a 
higher temperature for full sintering. 

The third goal is to determine the physical properties of monolithic fully 
dehydrated gel-silica glasses. In Chapter 5 various physical properties of the dense gel- 
silica glasses are compared with commercial melt/cast vitreous silica glasses (fused 
quartz) and other high-quality optical silica glasses (synthetic fused silica). 

The fourth goal of this study is to develop the technology for fabrication of 
transition-element doped xerogels. This is described in Chapter 6. Optical color filters 
that selectively transmit part of the visible spectrum can be made from xerogels doped 
with transition metal compounds. Transition elements, having unpaired electrons in 
their d-orbitals, can absorb light by ligand field-controlled transitions that do not 
involve variable valence states. The energy level scheme is controlled by the number and 
symmetry of the ligands and the strength of the ligand field [29]. The doped xerogels 
processed at different temperatures exhibit different densities and slight changes in 



11 



bonding strength which can produce a dramatic shift in their color response. For 
example, a 160°C silica xerogel containing 0.25% cobalt is a reddish-orange color, 
whereas the 850°C sample is a deep blue, and the 900°C sample has a greenish-black 
color. 

Finally, a summary (Chapter 7) is presented which reviews the present state of 
sol-gel processing science as applied to gel-silica optical monoliths and the properties 
of these unique materials. Questions still to be answered by future investigations are 
also included in the summary chapter. 



CHAPTER 2 
SOL-GEL TRANSFORMATION AND EXPERIMENTAL PROCEDURES 



Introduction 

During recent years many researchers have attempted to produce large monolithic 
dried xerogels; however, a reliable process had not yet been established at the time this 
work began [30-36]. Difficulties associated with this sol-gel processing method arise 
during all phases of aging, drying, and densification, clearly indicating insufficient 
understanding of basic changes in the ultrastructure during the sol-gel transformation 
and in the chemical reactions of the solvents, precursors, and catalysts involved. In 
general, crack formation during drying is a result of strong hydrostatic stresses within 
a relatively weak gel structure. Catastrophic failure can be avoided by adjusting the 
mechanical strength of the gel structure to exceed that of the hydrostatic force and/or by 
decreasing the hydrostatic stress relative to the gel's strength. 

The object of this chapter is to describe the principal mechanisms of the sol-gel 
method by which monolithic xerogels may be reliably produced. Four factors are used to 
describe the sol-gel transformation up to the gelation point: (1) the isoelectric point 
(iep), (2) the point of zero surface charge (pzc), (3) thermally activated particle 
movement (Brownian motion), and (4) Van der Waals force. Three kinds of dried 
monolithic gel samples were routinely prepared to aid in this study: pure silica, silica 
doped with transition elements, and silica doped with rare earth elements. 

Literature Review of Sol-Ge? Transformation Modeling 
Dr. Ralph K. Iler's pioneering work in the investigation of silica chemistry is the 
foundation of many of the ideas discussed in this chapter. Her found that silica gels can be 



12 



13 

obtained from supersaturated aqueous solutions produced by one of the following 
methods: 

(I) Concentrating an unsaturated silica solution by evaporating its solvent. 

(ii) Cooling a hot saturated silica solution. 

( i i i) Lowering the pH of an aqueous solution of a soluble silicate below 10.7. 

(iv) Hydrolyzing Si(OR)4 -- (where R is CH3, C2H5, or C3H7). 

In this study all of the monomers were produced by chemically hydrolyzing 
tetramethylorthosilicate (TMOS), as indicated in method (iv). The amount of monomer 
generated within a given period of time depends on temperature and the relative amounts 
of DCCA, water, and TMOS. When a solution of monomer, Si(OH)4, is formed at a 
concentration greater than the solubility of the solid phase of amorphous gel silica in 
water, and in the absence of a solid phase on which the soluble silica might be deposited, 
the monomers then polymerize by condensation to form dimers (two silicons), then 
tetramers (four silicons), then particles (eight or more silicons). For most alkoxide 
syntheses, a polymerization reaction occurs before hydrolysis is completed (as 
evidenced by 29 Si NMR studies [37, 38]). As shown in Figures 2-1 and 2-2, the 
particle's size at any moment of growth is controlled by the Ostwald ripening mechanism 
[see p. 175-220 in ref. 4] and essentially is determined by the pH of the DCCA/silicic 
acid solution. 

Vysotskii and Strazhesko [39] describe that in the presence of a given acid, the 
growth of monomers is governed by the chemical equilibrium kinetics of the sol and is 
minimized at the isoelectric point (iep). This implies that the monomers grow to some 
certain size before the solution reaches its own iep. The iep occurs when the net 
electrical mobility of surface ions on the silica particles is zero and at a pH at which 
there is no charge outside the hydroelectric slip plane (outside this plane the liquid is 
free to move, inside the plane the liquid molecules are held too tightly to move) [see p. 



14 



9 



H-O-Si -O-H 
I 



H HI 

6 O 

I I 

H-O-Si — O— Si — O— H 



s 



9 

H 



monomer 



O 

i 
H 

dimer 



H 
I 



H 
H-O-Si — O— Si— O 



S 9 

H — O-Si — O-Si — 0— H 
O O 



cyclic tetramer 



particle (less than 10 A) 




particle size 
smaller than 50 A 



particle size 
larger than 50 A 



Figure 2-1 Particle growth in solution 



15 



100 



90 



80 



70 ~ 



§> 60 

as 

I 50 

g 
To 40 

a 
o 
o 

o 30 
o 

X 

O 20 
W 



10 





amount of Si(OH) 4 
polymerized into particles 



gelation 
point 



relative time scale 



Figure 2-2 Polymerization reaction occurs before hydrolysis is completed 



16 

660 in ref. 4]. Below are equations related to particle growth under two different pH 
conditions, and will be described as two models in the following paragraph. 

< pH < 2, [H + ] as a catalyst 

Si n O a (OH) b + + Si(OH) 3 + OH" -> Si n O a (OH)b-iOSi(OH) 3 + H 2 (1) 

2 < pH < 7, [OH-] as a catalyst 

Si n O a (OH)b+ -OSi(OH) 3 + H+ --> Si n Oa(OH)b-iOSi(OH) 3 + H 2 (2) 
Si n O a (OH)b is a surface hydrolyzed silica particle, where "n" can be 2, 4, 8, 40, 311, 
1438, etc. [see p. 8 in ref. 4]. The number of anhydrous oxygens within a particle is 
represented by "a"; "b" is the number of surface hydroxyl groups per particle. 

In an extensive study of silica polymerization, Linsen, Okkerse, Vysotskii and 
Strazhesko [39, 40], found the iep to be between pH of 1.0 and 2.0. Condensation is 
slowest in this pH range, thereby producing a minimum gelation rate. Gelation occurring 
at the iep results in gel structures of maximum specific surface area and maximum 
strength. These structures occur because the rate of aggregation is minimal as is the 
growth rate of the ultimate particles from the monomer. Consequently, the ultimate 
particles are smallest when the gel is formed at the iep. 

Strong Acid Model 

Figure 2-3 represents experimental data of relative gelation time versus solution 
acidity found by many researchers [39, 41, 42]; the corresponding relative surface 
area curve is shown in Figure 2-4. These two figures show that the longest gelation time 
results in the highest surface area when the solution was prepared at pH=2. This is 
because the rate of polymerization reaction depends on a catalytic effect which is at a 
minimum at pH=2. From these data a model is developed describing the gelation 
phenomenon in a strongly acidic solution, in which the pH is less than 2.0. 

The very high hydrogen ion concentration at pH<2.0 results in a rapid reaction 
among monomers to form dimers, cyclic tetramers, and very small particles, producing 



17 



CD 

E 
+-■ 
c 
g 

IS 

CD 

© 

> 

is 

CD 
DC 




PH 



Figure 2-3 Relative gelation time versus solution acidity 



18 



1 1 _ 



05 
0) 

05 
0) 

o 

03 

t: 

13 
U) 

> 

05 




pH 



Figure 2-4 Relative surface area versus solution acidity 



19 

a significant amount of free water (equation #1) which dynamically reduces the 
hydrogen ion concentration. This dilution slows the reaction between monomer and the 
particle surface causing a build up of monomers around the particle while the total 
hydrogen concentration in solution is reduced. This causes the pH to be increased to the 
isoelectric point with a pH approximately equal to 2.0. This implies that the stronger 
initial acidic solution (pH < 2) allows the monomers to grow to relatively larger 
particles before the iep is achieved and results in a relatively smaller surface area, as 
measured. As soon as the iep is reached, the particle size is nearly determined, the slip 
plane of the electrical double layer is formed, no electric charge outside the slipping 
plane can be measured, and particles are then homogeneously distributed in the solution. 
As shown in Figure 2-5, the particle surface has a slight negative charge in the 
presence of the positively charged monomer (equation #1). The monomers confined 
within the slip plane of the electrical double layer [see p. 358-378 in ref. 4] will 
gradually react with the particle surface under the influence of the hydrogen ion 
concentration and thermal energy, resulting in slight particle growth. Free water is 
released, diluting the hydrogen ion concentration while particle growth decreases. As the 
confined monomers are consumed, the electrical double layer and slip plane is 
eliminated. Formation of an electrically neutral particle surface, referred to as the 
point of zero surface charge (pzc), at pH = 2.5 [41] marks the beginning of gelation 
under the influence of thermally activated Brownian motion and Van der Waals attractive 
force from this strongly acidic sol. 

Weak Acid Model 

The mechanism for gelation in a weaker acid solution (pH 2.0 - pH 7.0) is 
somewhat different from that of a strong acid solution, as shown in Figure 2-6. The 
reduction in hydrogen ion concentration effectively weakens its strength as an acid 
catalyst preventing the hydrogen ion from attracting the hydroxyl group from the 



20 



monomers expose their positively charged 
electric cloud toward particle surface. 




\ E (HO)!S+- 



> fr 1 7 



£>' 7S 



\£ *P 






electrical double layer 



Figure 2-5 Particles in strong acidic solution. pH<iep (~2.0). 



21 



particles surface exposes positively charged electric 
cloud toward negatively charged monomer in the weak 
acid solution. 



H ~°\ _.;.OSKOVV\ 
o-— -+h\ 







1 
ll 

o 



\ - ; ° / r 1 +H / 

% » +H • -^? 



,# + H 



o 



electrical double layer 



Figure 2-6 Particles in weak acidic solution. pH>iep (pH 2.0 - pH 7.0). 



22 

monomers around the particle and exposing the negatively charged ~OSi(OH)3 molecules 
(equation #2) which can react with the particle's positively charged surface. Rather, 
the negatively charged oxygen of the hydroxyl ions in solution can attract a hydrogen 
from the monomer. This forms free water and leaves the negatively charged oxygen as a 
site now available to react with the positively charged silicon on the particle surface, 
thereby regenerating this basic catalyst as a hydroxyl ion is released. With the 
production of free water, the hydroxyl concentration is reduced, decreasing the pH as 
well as the hydroxyl ion's ability to act as catalyst which causes a build-up of monomers 
surrounding the particle surface. As the concentration of these monomers within the slip 
plane reaches a maximum, at about pH = 2.0, the isoelectric point (iep) is attained. At 
this pH the hydrogen ion acts as a catalyst promoting the reaction between the monomers 
and the surface hydroxyl groups which facilitates particle growth. Free water is a by- 
product of this reaction, reducing the hydrogen ion concentration and increasing the pH. 
This process continues until the monomer concentration inside the slip plane is 
exhausted and the electrical double layer eliminated. Thus, the point of zero surface 
charge (pzc) has been reached and gelation begins under the influences of thermally 
activated Brownian motion and Van der Waals attractive force. 

Brownian Motion. Van der Waals. and Interparticle Bonding Models 

When monomers come together to form very small (10 A - 50 A) [43], uniform, 
uncharged (pzc) particles, their motion is essentially governed by thermal diffusion as 
described by the diffusion equation below. 

D = KT/(3jiti d) (3) 

where 
D is the diffusion coefficient 
T| is viscosity 
d is the effective instantaneous diameter of the polymerized cluster 



23 

K is Boltzmann's constant 
T is absolute temperature 
The average displacement X of a particle from time zero (@ pzc) to any point in 
time t is: 

X = (2Dt)1/2 (4) 

Prior to reaching the pzc, the viscosity of the sol increases only slightly, as shown in 
Figure 2-7 [44]. At the point (pzc) is achieved the small particles are homogeneously 
distributed throughout the solvent, as shown in Figure 2-8. Governed by Brownian 
motion (equation #4), these thermally activated, hydroxyl ion-catalyzed particles 
randomly collide under the aid of Van der Waals attractive forces and a base catalyst, as 
shown in Figure 2-9, to form long spherical-particle chains. As these chains continue to 
form, the viscosity increases until there exists a three dimensional network throughout 
the volume of the sol, as shown in Figure 2-10. This is described as the gelation point. A 
sol takes a specific time to reach its own gelation point. 

Gelation time is then defined as at the moment the sol is prepared to the moment the 
sol loses its freedom to move. The length of gelation time is a function of the temperature 
and the relative amounts of acidic DCCA, water, and TMOS in a sol. Figures 2-11, 2-12, 
2-13, and 2-14 show that the gelation time can be exponential curve fitted with one of 
the four variables (i.e. temperature, oxalic acid (DCCA), water, and TMOS) in which the 
other three are kept constants. Increasing the sol temperature promotes the thermally 
activated Brownian motion and thus decreases the gelation time as shown in Figure 2- 
11. A decreased amount of oxalic acid concentration weakens the catalytic effect among 
particles and therefore increases the gelation time as shown in Figure 2-12. An 
increased TMOS content in water results in an increased concentration of particles and a 
decreased distance between particles, which consequently, shortens the gelation time as 
shown in Figures 2-13 and 2-14. 



24 



V) 

o 
o 

w 

"> 

> 

TO 

0) 













c 
o 

o 

E 






p disappearance 
•gion 


Brownian 
region 




iep formation region 


.9 ȣ 


JPZC 

i 



tg 



relative time scale 



*t n is the gelation time 



Figure 2-7 Relative viscosity versus time 



25 




Figure 2-8 Homogeneous particle distribution throughout the solvent. 



26 



(a) Brownian motion and Van der Waals forces 




(b) base catalyst 




(c) vacancies are created 
in the neck area 



[OH"] 




° H °H HH hH 



Figure 2-9 Particles collide randomly with the help of Van der Waals 
attractive forces, Brownian motion and base catalyst. 



27 



aaaj|- 



100 A 




Figure 2-1 Acid catalyzed particles constitute fibrillar chains throughout 
the volume of sol. 



28 



I 

E 
© 



3000 



2500- 



2000- 



1500 



1000- 



500 



20 



Dl water 400 cc 
oxalic acid 8 grams 
TMOS 200 cc 




60 80 

Temperature (°C) 



100 



Figure 2-11 Gelation time versus temperature. 



29 



1200 



1000- 



800- 



0) 
•I 600 H 

£Z 

o 

1 

<§ 400 



200 - 



TMOS 200 cc 
Dl water 400 cc 
temperature (°C) 




ft.- ' 



ii i i i i ii — i — i- 



-i — ■ — i i — i- 



10 



15 



oxalic acid (gram) 



Figure 2-12 Gelation time versus oxalic acid content. 



400 



30 



300- 



1 

E 

c 
o 

1 

o 



200- 



TMOS 200 cc 
oxalic acid 8 grams 
temperature 55°C 



100 - 



100 200 300 

Dl water (cc) 



— i — 

400 



500 



Figure 2-13 Gelation time versus water content. 



31 



c 
1 

CD 

E 

■*-• 

c 
<2 

I 

© 



800 - 
600- 


— f— 


Dl water 400 cc 
oxalic acid 8 grams 






temperature 55°C 


400- 














\ l 


i 


200 - 










^vj 


! 


0- 


' 1 ■ 1 


r — ■- ■ - 







100 



200 



300 



400 



500 



TMOS (cc) 



Figure 2-14 Gelation time versus TMOS content. 



32 

Characterization of Gelation 

Professor Paul Flory's theory of gel formation [45, 46], with which Her agrees 
[see p. 176 in ref. 4], notes that the silica monomer has four polymerization functional 
groups (f=4). The degree of polymerization (DP) obtainable in a system is therefore 
described by the equation: 

DP = 1/(1 -pf/2) (5) 

in which "p" is the percentage of reacting monomers (that is the fraction of the total 
concentration of monomer which is the reaction product from TMOS) and T is the 
number of polymerization functional groups. At the gelation point the degree of 
polymerization approaches infinity, therefore (1 -pf/2) must equal zero. For f=4 the 
percentage of total concentration of monomer going into gel phase must equal 50%. Since 
equal amounts of monomer exist in the liquid as well as in the gel, no refractive index 
change is observed at the gelation point. Consequently, the xerogel remain optically 
transparent throughout gelation. 

Aging Mechanism 

Aging is a process by which the gel structure is reinforced via surface area 
minimization of the spherical particle chains; this is shown in Figure 2-15. The surface 
area can be minimized by four possible mechanisms: (1) condensation of surface silanol 
groups (zipper effect) which creates stress and then results in vacancies in the neck 
area between particles, (2) thermally activated transportation of silica molecules from 
the volume or from the particle neck boundary to vacancies, (3) deposition of monomers 
from the liquid into the negative curvature area of two weakly connected spherical 
particles, and (4) dissolution of monomer from the particles' area of positive curvature 
into the pore liquid, as shown in Figure 2-16. 

The first, third, and fourth mechanisms do not result in gel shrinkage; the second 
of these mechanisms does [47]. The particle rearrangement involved in the second 



33 



(a) No surface area minimization at the time 
of gelation point. 




(b) surface minimzed after aging. 




d - d, + d 2 



Figure 2-15 Surface minimization in the neck area. 



34 



(a) at the time of gelation point (t g =0) 




(b) the first mechanism: formation of 
vacancies in the necks of chain, the total 
length d does not change 




(c) the second mechanism: migration of vacanices 
from neck area out of gel body, the total length d 

shrinks. 




(d) silanol groups depart from the positive 
curvature area of particle's surface (the third 
mechanism) and deposit on the negative curvature 
area (the fourth mechanism). 




monomers 
deposit 



silanol group depart 



Figure 2-16 Surface minimization during aging. 



35 

mechanism is initiated by thermal energy. Therefore, the higher the aging temperature, 
the faster is the rate of matter migration to vacancies and the more rapid is gel 
shrinkage, as shown in Figure 2-17. About the same maximum shrinkage (=28%) is 
associated with each aging temperature. It is possible that the same amount of vacancies 
are quickly created inside the necks between particles during the first stage, mechanism 
No.1 , for all identical gels. Subsequently, all of these vacancies are annealed out of the 
gel body in the second stage, mechanism No. 2, and then equal shrinkage is obtained. The 
same maximum shrinkage in the aging stage is probably predetermined by the 
processing characteristics of each gel (e.g. pH, water, DCCA, TMOS ratio). The gel 
shrinkage kinetics can also be monitored by the time at which 28% maximum gel 
shrinkage is observed at each temperature, as shown in Figure 2-18. Shrinkage 
improves gel strength; therefore, a relatively hard and dense gel can be obtained as a 
result of optimizing the aging process. Figure 2-19 shows the increase in gel 
microhardness with percentage of shrinkage. It is this increase in mechanical strength 
with aging that makes it possible to obtain dried monolithic xerogels. 

Drying Modeling 

Control of drying is critical; without a full understanding of the gel's drying 
mechanism and the development of a suitable method to deal with it achieving a dried 
xerogel without cracking is very difficult. Drying control involves both chemical and 
physical aspects. Chemically, the use of an acidic DCCA in the sol minimizes the particle 
size which results in an increased gel strength and a more homogeneous particle-size 
distribution, thereby diminishing uneven pore stresses. Physically, the use of a drying 
control chamber decreases the effect of differential pressures on the gel body which 
could lead to stress fracturing. 



36 



40 



30 



CD 

03 

c 



CO 

© 
E 

o 

> 



c 

CD 

o 

a> 
a. 



20 



10 - 




30 60 90 

time (hr) 
*time starts from the gelation point. 



120 



150 



Figure 2-17 Shrinkage of silica gel inside 100 cc polystyrene cylinder 
as a function of aging time. 



37 



80 



60 



O 
2 40 

CO 

i— 

a> 

CL 

E 
i- 



20 _ 




30 



60 90 

time (hr) 



Figure 2-18 The time silica gels shrink to 72% of original volume versus 
aging temperatures inside 100 cc polystyrene cylinder. 



38 




10 15 

Percent of shrinkage 



Figure 2-19 Microhardness of aged gel versus percentage of shrinkage. 



39 

A silica gel is defined as "dried" when the physically adsorbed water is 
completely evacuated and no significant weight loss is observed at increased 
temperatures. 

Cracking during the drying process is essentially the result of differential 
evaporation of pore liquid, Figure 2-20, as discussed in detail by J. Zarzycki [48]. The 
Laplace equation is used: 

AP V | = P| - Pv = 2y v l COS8/R (6) 

where AP V | is the differential capillary vapor pressure between the surface of the vapor 
phase (in which vapor pressure = P v ) and the liquid phase (in which vapor pressure = 
Pi), within a very small pore of radius R. In equation (6) Yvl is the specific surface 
energy, and G is the contact angle. 

Theoretically, to prevent shattering of the gel body during drying, the capillary 
vapor pressure in the liquid phase (which is transmitted to the wall of the pore channel) 
must be offset by the capillary vapor pressure in the vapor phase. For AP V | to equal 
zero, the cosine of the contact angle must also equal zero (@ 9=90°), as the radius (R) 
and the surface energy Cm) at the liquid-vapor interface will always have some value. 
Young's equation for the equilibrium of a solid (s) - liquid (I) - vapor (v) system is 
derived by balancing the horizontal components of the specific surface energies, Ysl. 
Ysv. Yvl of the system. The equilibrium equation is given as: 

Ysv = Ysl + Yvl cos e (7) 

As cosine 9 becomes zero, this equation simplifies to ysv = Ysl- This means that the 
work required of the liquid to act on the wall of the solid is the same as the work 
required of the vapor to act on the wall of the solid. 

cos9 = o 

Ysv = Ysl 

A S v =A S | 

YsvdA sv =Ys|dA s | 



40 



stress initiated crack lines 



gel body 




pore liquid 



Figure 2-20 Differential evaporation. 



41 

PsvdV sv = dw sv = YsvdAsv 

PsldV S | = dw s | = YsldA S | 

AP v |=Psv-Psl=0 

PV = nRT (for ideal system) 

PV = W/M XT (for real system) 

PsvdV sv = XT sv d(W/M) = P S |dV S | = XT sv d(W/M) 



where 



W = weight of vaporized liquid 

M = molecular weight of liquid 

X = vapor constant 
The actual pressure of the vapor phase per unit area of pore wall is the same as the 
actual pressure of the liquid phase per unit area of pore wall; this is called the 
saturation point or equilibrium vapor pressure [49]. When AP V | is zero, there is no 
difference in the liquid level within the capillary pore channels regardless of the pore 
radius, as shown in Figure 2-21. 

However, for the case of drying actual xerogels the differential pressure AP V | can 
be minimized to zero with the use of a proprietary device. This device keeps the vapor 
pressure in the vapor phase, P v , at a value the same as that of the vapor pressure in the 
liquid phase, P|, in the gel. As a result, AP V | is zero and gel remains intact. The vapor 
pressure within this device is controlled by the temperature which must be carefully 
maintained. At temperatures higher than the boiling point temperature of the gel pore 
liquid, the vapor pressure in the liquid phase, P|, exceeds one atmosphere (P v will 
never be higher than 1 atm in this device because it is not an autoclave system). Thus 
the system will equalize as gas escapes from the device, i.e., AP V | is not zero, which 
would cause a differential vapor pressure AP V | between the liquid and the vapor phases 
sufficient to shatter the gels, as shown in Figures 2-20 and 2-22(a). 



42 



gel structure 



pore 




Figure 2-21 No differential evaporation. 



43 



(a) 



P(air) m one atmosphere (1 atm) 



vapor out 




• * 'il'v^'"'^"! J* 

.■.V*-V ....; 
" "J. ' I ''"j ■'. • ,'*£{&• 



■■■•■•. :>•'*>•. fl&r 




P(vapor) > P(air) 



(b) 



P(air) = one atmosphere (1 atm) 



air in 



air in 




P(air) > PCvapor) 



Figure 2-22 Gel cracks inside nonequivalent evaporation containers. 



44 

At temperatures lower than the boiling point of the gel pore liquid the vapor 
pressure in the liquid phase (P|) is less than one atmosphere; therefore, air will enter 
the device to establish a vapor phase pressure (P v ) equal to 1 atm, resulting in a 
differential vapor pressure (AP V |) which is not zero (Figures 2-20 and 2-22(b)). 
However, by maintaining a zero differential pressure the capillary force is eliminated 
(Figure 2-23), thereby significantly removing the differential hydrostatic stresses 
within the gel body and retaining the gel's monolithic shape. 

Structural Characterization 

The gel consists of a three-dimensional network of silica particles rigidly linked 
together. If the structure of the gel is relatively coarse, the gel body is fragile and likely 
to shatter. If the structure of the gel is relatively fine, consisting of fibrillar chains of 
very tiny particles, and therefore somewhat flexible, the gel will be strong enough to 
shrink considerably without cracking. However, the shrinkage of a silica gel is 
irreversible. Shrinkage occurs as the gel dries due to the surface tension of the liquid 
within the pores. As drying occurs it is probable that certain bonds on the necks between 
particles break, which allows portions of this area to be dissolved into the pore liquid 
and transported to areas of negative curvature, as shown in Figure 2-24. This is because 
solids minimize surface area so as to reduce surface energy to a minimum. 

An equation relating the solubility of a curved solid surface in water to the radius 
of curvature was derived by Ostwald and Freundlich [see p. 50-51 in ref. 4]: 

log(Sr/Sj) = KE/Tr (8) 

where S r is the solubility of a particle having a radius of curvature r; Si is the 
solubility of a flat surface with a radius of curvature of infinity in that water; E is the 
surface energy of the solid; T is the temperature; and K is Boltzmann's constant. The 
meaning of this equation is schematically illustrated by Her in Figure 2-25. As 



45 



P(air) = one atmosphere (1 atm) 




p - p — p 

r atm(air) ~ r vapor ~ r liquid 



Figure 2-23 Situation to avoid cracking. 



46 



shear stress AP 



silica fibrillar structure 




shear stress AP 



monomers depart and 
old bond broken 




monomers deposit 
and new bond formed 



Figure 2-24 Redeposition of monomers from the broken neck area to the area of 
negative curvature. 



47 




:::: 



WdM 



SAy.-y&ym ■ ■:■: :^y:y:^-yy.^y.o:^yy-y, .-:■:■, ■:■;■:-'■:■:■ -yyy ■:■:■■ ■■■:■::■:■: 

>y;:;:;:;:;: : :; : :;:;:;:-:-:y:;:^^ 



tK-S^»:2X*:-: 



f'x5j£i!5i : ^siii „ 



increasing negative curvature 








increasing positive curvature 



Si0 2 solubility, ppm 




-5 5 

diameter of curvature - nanometers 



Figure 2-25 Solubility of silica in neutral water at 25°C varies with the radius of 
curvature of the surface according to the Ostwald-Freundlich eqiation. 



48 

shown, when an acidic silica gel is sufficiently dried to contain pores (negative 
curvature in left side of the figure) that are only a few nm in diameter a small decrease 
in pore size results in sudden elimination of the pores. Table 2-1 confirms that the 
remaining uniform pores stay unchanged in diameter but decrease in total volume and 
surface area as the sintering temperature increases. 

A mechanism was suggested by Her [47] that in a densification process, gel 
shrinkage is the results of sudden decomposition of pores into vacancies in the gel 
structure and traveling vacancies which migrate to the outside of the gel body along the 
surface of the pore network and do not remain in the pores to enlarge them [50]. 

Experimental Procedure 
Large scale monolithic dried silica gel samples (up to 10 cm x 8 cm x 2 cm), as in 
Figure 2-26, have been routinely produced by applying the concepts and mechanisms 
stated in Section II of this chapter. Several kinds of standardized samples were made for 
characterization in this study including pure silica gels, cobalt- copper- and nickel- 
doped silica gels, neodymium- and erbium-doped silica gels. The two examples described 
below detail the procedure used to produce both pure silica and doped silica samples. Six 
steps are generally needed to produce the sol-gel derived monolithic silica gel-glass 
samples, as shown in Figure 2-27. The drying control chemical additive (DCCA) is 
introduced in Step 1 ; this makes it possible to control each of the five subsequent steps 
and prevent gel shattering. 



49 



Table 2-1 
Oxalic acid (5.0 grams) as DCCA in 200 cc H2O/IOO cc TMOS 



Temperature 200°C 450°C 750°C 800°C 830°C 

Surface area 

(m2/g) 651.12 612.10 413.25 385.47 335.40 

Total pore volume 

(cc/g) 0.36 0.33 0.22 0.20 0.18 

Average pore radius 

(A) 11.02 11.03 11.06 11.03 11.05 



50 




Figure 2-26 Picture of a large scale 160°C dried silica gel sample. 





Step 1 




Step 2 












Step 3 








Step 4 




Step 5 




Steps 








Figure 2-27 Procet 



51 



Mixing 
(H 2 + DCCA + TMOS) 



Shape formation 
(casting) 



Sol-gel transformation 
(gelation) 



Aging 



Drying 



Densification 



52 

Example Qne; 

Production of dried pure silica gel monolith from oxalic acid DCCA 

Step 1: Mixing 
Tetramethylorthosilicate (TMOS) is used as a precursor for silica monomers to 
form Si-O-Si bonds in the gel structure. The mixing of water with TMOS forms a silica 
sol via the following simplified hydrolysis and polymerization reactions: 
Si(OCH 3 ) 4 + 4 H2O — -> Si(OH) 4 + 4 CH3OH 

-Si-OH + OH-Si > -Si-O-Si- + H2O 

The specific standard procedure followed in Step 1 is: 

(a) Pour 300 cc of water into a clean 800 cc beaker. 

(b) Place the beaker on a hot-stirring plate. 

(c) Mix 6 grams of oxalic acid with water using a PTFE coated magnetic bar; 
control via the hot-stirring plate. 

(d) Stir for 5 minutes to get a homogeneous solution. 

(e) Add 150 cc TMOS to the acid solution, while continuing to stir 
vigorously for approximately 50 minutes. 

(!) Immediately increase the temperature from 25°C to 85°C by raising the 
temperature on the hot-stirring plate to maximum. 

> ' " w3^ ;."- t #3 

(g) If feasible, carefully place ice water in a three-layer polystyrene 
thin film on top of the beaker to condense the hot vapor and return 
it to its solution. 

(h) Continue stirring and heating for approximately 50 minutes before 
casting. 



53 

Step 2: Casting • 

(a) The intimately mixed sol is cast from its heated vessel into a mold (20 mm H x 
100 mm D) that corresponds to the final desired shape. For best surface results, 
polystyrene is the selected mold material. 

(b) The duration of the casting operation is not critical since gelation does not 
occur until after casting is completed. 

Step 3: Gelation 
Gelation occurs in the mold with the resulting solid object taking the shape and 
surface finish of the mold. Gelation times with oxalic acid are typically 20 hours at 25°C 
and 4 hours at 70°C, depending on the relative concentrations of water, TMOS, and DCCA, 
as shown in Figures 2-11, 2-12, 2-13, and 2-14. 

Step 4: Aging 
The solidified gel is then placed into an aging oven at a temperature ranging from 
50°C to 80°C for a times ranging from 20 to 48 hours to achieve maximum shrinkage. 

Step 5: Drying 
Prior to Step 5, control of the gel ultrastructure is governed by the DCCA which 
allows removal of the pore liquid without cracking the gel. Typically this is done by first 
removing the excess liquid present after gel shrinkage in Step 4. The pore liquid is then 
removed, consistent with the theory stated in Section II of this chapter, by confined 
evaporation over a temperature range from 70°C to 160°C for times ranging from 18 to 
90 hours. An example of a typical heating program is shown in Figure 2-28. 



54 



200 




60- 
40- 

20 



1 



— I ' 1 ' 1 > ! ■ 1 ■ 1 ' 1 •"" 

25 50 75 100 125 150 175 200 



Time (hr) 



Figure 2-28 Drying program for wet gel. 



55 

Step 6: Densification 
The ultraporous dried silica gels are converted to partially dense monoliths by 
heating from 150°C up to 900°C over a period of 3 to 6 days; samples are taken out of 
the furnace at the end of the heating program. An example is shown in Figure 2-29. 

Example two: 

Production of dried transition and rare earth element doped silica gels from nitric acid 
DCCA. 

Step 1: Mixing 

(a) Add 60 cc (1 N) HNO3 (nitric acid) to 340 cc of distilled water at room 
temperature and mix for 5 minutes with a magnetic stirrer. 

(b) Add 200 cc TWOS to the nitric acid water solution while continuing to mix 
vigorously, increasing the solution temperature to 85°C for no more than 60 minutes. 

Step 2: Casting 
The intimately mixed sol (60 cc) is cast from its heated vessel into a polystyrene 
mold (20 mm H x 100 mm D) at room temperature. The length of time for casting 
should be no more than 110 minutes since gelation will take place during prolonged 
casting operation. 

Step 3: Gelation 
Gelation occurs in the mold at 55°C in 115 minutes with the resulting solid object 
taking the shape and surface finish of the mold. 



56 



1000 



800 



o 

o 



E 600- 



© 

Q. 

E 



400 




200- 



75 100 125 150 



Time (hr) 



Figure 2-29 An example of a silica gel-glass densification program. 



57 

Step 4: Aging 
The solid is aged in the mold initially at 55°C for 10 hours, followed by an increase 
to 80°C for 15 hours. 

Step 5: Drying 
The aged pure-silica gel is removed from the mold and dried with a controlled 
evaporation rate, as described in Section II of this chapter, initially at 70°C, gradually 
increasing the temperature to 160°C during a 90 hour period. 

Step 6: Impregnation 

(a) One gram-percent of transition metal element (i.e., cobalt nitrate, nickel 
nitrate, copper nitrate) or three gram-percent of rare earth element (i.e., neodymium 
nitrate, erbium nitrate) in deionized (Dl) water is prepared for doping, or 
impregnating, the completely dried gel. The dried gel is immersed into the solution, 
whereby the interface between the liquid and the voids migrates from the exterior into 
the center of the gel body in the rate of 0.5 cm/hour, as shown in Figure 2-30. 

(b) The doped gel is then placed in the drying oven at 200°C for 12 hours to 
remove the pore solvent. 

Step 7: Densification 

(a) The fully dried silica gel doped with transition metal or rare earth elements is 
heated to 400°C to eliminated any residual nitrates via conversion to its gaseous oxides. 

(b) Additional densification can be achieved by heating from 400°C to 1000°C. 

Results 
Monolithic samples of pure silica gel, transition metal element doped silica gel, and 
rare earth element doped silica gel were routinely produced following these procedures; 
some are shown in Figures 2-31 to 2-35. The physical and optical properties of these 
samples will be discussed in succeeding chapters. 



58 



top view 




Figure 2-30 Sample immersion into transition metal or rare earth nitrate/water solution. 



59 



K8 



J 



\ 



Figure 2-31 Picture of a 160°C dried silica gel. 



60 







Figure 2-32 Picture of a cobalt nitrate-doped silica gel which was stabilized at 

750°C and redried at 160°C. 



61 



M14 




Figure 2-33 Picture of nickel nitrate-doped silica gel which was stabilized at 750°C 

and redried at 160°C. 



62 




Figure 2-34 Picture of copper nitrate-doped silica gel which was stabilized at 750°C 

andredried at 160°C. 



63 




Figure 2-35 Picture of neodymium nitrate-doped and erbium nitrate-doped silica gels 
which were stabilized at 750°C and redried at 160°C. 



64 
Conclusions 

It was not found necessary to add methanol to this xerogel system, though this is the 
practice of many researchers [51-56]. 

Addition of oxalic acid or nitric acid as DCCAs is necessary in the mixing step of 
both Examples #1 and #2 as an acidic DCCA controls the radius of the individual silica 
particles to a few nanometers that form during the early stage of monomer growth and 
the subsequent fiber-like polymerization. 

The particles are made uniform due to Ostwald ripening at any moment of growth. 
As soon as particle growth stops at the pzc (point of zero surface charge), an electrically 
neutral particle surface forms; therefore, thermally activated Brownian motion, Van 
der Waals attractive forces and base catalytic effects among particles in the sol become 
the driving forces to form particle chains which reach the gelation point. 

During aging, the reinforcement and the shrinkage of the fibrillar network of a gel 
proceeds as a result of growth of interparticle necks and migration of vacancies to the 
exterior of the gel. The rate of aging shrinkage is primarily determined by the rate of 
thermally activated vacancy migration. 

After aging, the interparticle necks comprise a very large fraction of the gel 
fibrillar structure and become relatively flexible (like glass fibers are flexible). 
Consequently, the gel can endure certain hydrostatic stresses and shrink considerably in 
the drying stage without cracking, as illustrated in Figure 2-36. 

Differential vapor pressure (AP V |) is the stress which shatters the relatively 
weak gel into pieces in the drying stage. A gel can be dried without cracking by using a 
drying device which eliminates the differential vapor pressure between vapor phase 
(P v ) and liquid phase (P|) inside the capiiiary pores. 



65 



M, 




stress 

in presence 
of water 



rigid gel structure 




broken 
neck 



M 1 =M 2 



M, 



stress 




flexible gel 
structure 



Figure 2-36 Fibrillar gel structure is relatively flexible compare to coarse gel structure. 



66 

Monolithic gels with an optimal ultrastructure and high resistance to drying 
stresses, which are chemically controlled (by adding acidic DCCA) and physically 
stabilized (by introducing a drying device), are identified by a change in visible light 
scattering during the drying process. The optical sequence for a drying gel is as follows: 
complete transparency with a very slight blue tone, followed by an opaque stage, 
followed by transparency. These changes in optical properties can be used to monitor the 
drying process and therefore offer the potential to be used in a feedback loop to optimize 
drying. Monitoring weight loss can also help to achieve the final stage of drying; when 
the theoretical molecular weight of silica is reached, drying is finished. This process can 
be automated and used with computer aided processing. 

The fully dried gels can be modified by liquid phase impregnation of various 
chemical species (e.g., compounds of transition or rare earth elements) into the dried 
gel. Because of the extremely small size (10 A - 100 A) of the ultrapores in the gel, it 
is possible to introduce a very homogeneous ion distribution within the gel matrix. For 
measurements the physical properties presented in Chapter 3, the ultraporous dried 
silica gels are converted to partially dense monoliths by heating from 150°C to 900°C 
over times ranging from one day to one week. 



CHAPTER 3 
PHYSICAL PROPERTIES OF PARTIALLY DENSIFIED SIUCA XEROGELS 



Introduction 

Monolithic, noncrystalline, dried xerogels of pure silica, hereafter simply called 
gels, have been made by the procedure stated in Example #1 of Chapter 2. These samples 
are heated to 150°C (the temperature at which the gels are free from physical water) to 
become standard dried gels. 

The physical properties of the fully dried gel are a function of the internal 
structure which depends on the various chemical and physical conditions during every 
step of processing (i.e., the relative amounts of water/DCCA/TMOS, temperature, 
pressure, and time for aging and drying). 

At sufficiently high temperatures thermal energy provides the driving force for 
ultrastructural rearrangement which decreases surface area and thereby minimizes 
surface tension inside the gel structure. This is the primary mechanism for 
densification [see p. 469-490 in ref. 23]. 

A large reduction in pore volume is accompanied by the decomposition of residual 
organic compounds into carbon dioxide (between 250° and 450°C) and also by the 
combining of surface hydroxyl groups resulting in some degree of dehydration. Both of 
these phenomena may cause thermally induced stress fracturing in the densification 
stage. However, by controlling the rates of these reactions silica gel monoliths that are 
crack-free, partially densifiecf and shrunk, can be successfully made at various 
temperatures, ranging from 200°C to 850°C. 

This chapter presents a study of the physical properties of partially dense silica- 
gel monoliths. Data were obtained from numerous measurements including structural, 
optical, thermal, and mechanical testing. Structural information was provided by 

67 



68 

Fourier-transform-infrared (FTIR) spectroscopy, ultraviolet-visible-near-infrared 
spectroscopy (UV-VIS-NIR), N2 adsorption-desorption isotherms interpreted using 
Brunauer, Emrnett, Teller (BET) analysis which includes specific measurements of 
surface area, pore size distribution, pore volume, and pore radius, as well as large angle 
X-ray diffraction. Optical information was obtained solely using an index of refraction 
test. Thermal data were collected from differential scanning calorimetry (DSC), 
differential thermal analysis (DTA), thermogravimetric analysis (TGA), and 
thermomechanical analysis (TMA). Mechanical properties (gel strength) were 
determined using flexural strength, compressive strength, microhardness, fracture 
toughness and density measurements. 

Review of the Literature 

Three mechanisms of densification are summarized by Zarzycki, et al. and 
Brinker, et al. [25, 57]: (1) polymerization reactions which serve to crosslink the 
network and partially release the surface hydroxyl groups, thereby forming free water; 
(2) structural rearrangements that occurs when segments of interparticle necks are 
broken and other neck segments become connected; and (3) viscous sintering 
accompanied by the combination of surface hydroxyl groups. The first two mechanisms 
cause a slight density increase; the third mechanism is a result of high temperature 
viscous flow which eliminates the pores so that the bulk density approaches that of fused 
silica. No gel can be completely dehydrated and converted into a fully dense glass (i.e., 
without foaming) in an ordinaiy air-atmosphere furnace; but fortunately, the gel can be 
partially sintered So a desired temperature, below the foaming point, and cooled to room 
temperature while remaining intact. 

Any material can give rise to absorption or emission of radiation within the 
allowed transitional, vibrational, and/or rotational energy levels. Infrared spectroscopy 



69 

(FTIR) can provide vibrational information on changes occurring in the gel structure 
during sintering [58]. 

Water terminates the bridging silicon-oxygen-silicon bonds on the particle's 
surface inside the porous gel, as shown in Figures 3-1 and 3-2. Water's disruption of 
the Si-O-Si bridging bond is similar to that of sodium ions within a dense soda silicate 
glass. This gives rise to absorption in the ultraviolet (UV) region of the optical 
spectrum. The UV-VIS-NIR spectra technique is an easier and more sensitive tool than 
the infrared method for understanding the evolution of bonding and identifying the 
species inside the gel structure in the densification process [59]. 

The measured surface area, obtained from BET analysis, of a standard dried gel is 
about 750 m 2 /g at 200°C. The particle size is calculated from Havard, Wilson's model 
[60] where the diameter is equal to a constant (2750) divided by the surface area. The 
particle diameter for a gel made by Example #1 in Chapter 2 is 3.6 nm at 200°C. The 
measured surface area is somewhat less than actual since nitrogen molecules, used in the 
BET analysis, cannot completely penetrate the negative curvature area between all the 
connected particles. However, the BET surface area measurement also includes the 
surface hydroxyl groups which increases the particles' measured surface area value; 
this increase is less significant than the decrease resulting from incomplete nitrogen 
penetration. 

Silica gel is essentially a special form of porous glass. Previous x-ray diffraction 
studies by Mozzi, Warren, Uhlmann and Wicks [22, 23] have established in detail the 
tetrahedral bonding arrangements in vitreous silica. The maximum in the distribution of 
Si-O-Si angles in amorphous silica is at 144°, with most angles being within 10% of 
this maximum. There is no evidence for a preference in fused silica for edge-to-face 
sharing of tetrahedra, which is often found in crystalline silicates. X-ray diffraction 
patterns generally exhibit a relatively broad peak for gels indicating the absence of 
atomic periodicity or long-range structural ordering compare to that of quartz. 



70 






pore 






water lii 

^ :■;-;■:■:■:■:■:■:-:■:■:■:■:■:■:■:■:-■.' V:' 1 >■:■'-'■. 


/ 


^V*^x \ 




^/ 






V ^ 






'''■■ : S;:;:||p' 




\ 


/ 





cut-off profile magnified in Figure 3-2 



Figure 3-1 Random sampling profile of gel skeleton. 



71 



A profile of gel skeleton 




Oxygen atom Q 
Silicon atom • 



Ge! fibrillar structure cut off profile 



Figure 3-2 Water terminates the Si-O-Si bridging bond on the particle's surface. 



72 

Consequently, a random edge-to-edge sharing of silica tetrahedra with variable Si-O-Si 
angles described above is proposed for silica gel fibrillar structures. 

The magnitude of index of refraction (n) indicates the extent of change of the speed 
of light by the electromagnetic field of a transparent dense material. The index of 
refraction can be expressed by Snell's law n( g | a ss)/n(vacuum) = sin e( V acuum)/sin 
9(glass) = V( Va cuum)/V{giass). where n(gj a£ s). v (glass). and 8( g iass) are the refractive 
index, the velocity, and the angle of refraction of glass respectively, n( V acuum), 
v(vacuum) are constants, and 9( V acuum) is the angle of incidence of light in vacuum. 

Index of refraction is a dependence of (1) the density, (2) the polarizability of the 
glass, and (3) the wavelength (X) of monochromatic radiation [61]. In this chapter 
partially densified silica gels are discussed where the chemical compositions are 
essentially SiC>2 and chemical bonded surface -SiOH groups. The nonbridging hydrogen 
ions (H + , a proton) of these silanoi groups contribute very little effect on oncoming 
light [see p. 660 in ref. 23], thus, the polarizability of these partially densified silica 
gels can be assumed to be a constant. Consequently, the variation of refractive index with 
density described by the Lorentz-Lorenz equation [see p. 658 in ref. 23] can be 
simplified as will be discussed in the Results and Discussions Section of this chapter. 

Differential scanning calorimetry (DSC) is used to measure the temperatures 
associated with transitions in materials, including boiling points, melting points, 
liquid-crystal transitions, heats of reaction, specific heat capacity, oxidative and 
thermal stability, purity, glass transitions, and reaction kinetics. 

Differential thermal analysis (DTA) gives the same qualitative information as DSC, 
but is used primarily for studies involving high temperatures which exceed the range of 
the DSC cell (700°C). 

Thermogravimetric analysis measures weight change as a function of temperature, 
and provides derivative TGA data used to quantify the chemical changes in a gel during 
thermal processing. 



73 

Thermomechanical analysis (TMA) measures the thermal expansion coefficient, 
glass transition temperature, softening temperature and provides data for gel shrinkage 
analysis [62]. 

Flexural (FLEX) and compressive (COMP) tests are performed to determine the 
material's strength under external mechanical loads. 

A Vickers microhardness test, which yields a value for the diamond pyramid 
microhardness number (DPN), is used to measure the mechanical resistance of a gel and 
gel-glass to diamond pyramid plastic indentation in a microscopic area of the surface 
[63]. The fracture toughness is obtained directiy from the crack length which extends 
outside the diagonal of diamond pyramid indentation during the Vickers microhardness 
measurement [64]. 

Bulk density measurements are used to monitor the change in gel structure during 
sintering; it also gives useful information for interpreting variations in refractive 
index. 

Experimental Procedure 
Samples, fabricated by the procedure stated in Example #1 of Chapter 2, were 
heated to various programmed temperatures in an ambient air furnace, as shown in 
Figure 3-3. The following tests, listed in Table 3-1 , were performed on these samples. 
The infrared spectra were recorded on a Nicolet MX-1 FTIR spectrometer 
equipped with a diffusion reflection stage and a microcomputer for data storage. The 
diffusion reflection stage in which the infrared passes into the bulk (about 0.5 mm deep 
and 20 mm 2 area) of the gel, undergoes reflection, refraction, scattering and absorption 
in varying degrees before returning back at the sample surface. The radiation reflected 
out from the gel is distributed in all directions of the surrounding hemisphere and 
corrected to form spectra by a highly reflective semispherical mirror. Chemical species 
and bonding information can be interpreted in terms of the position and intensity of IR 



74 



1000 -, 




Time (hr) 



Figure 3-3 Heating programs for various samples 



75 



Table 3-1 
Physical property measurements 

TEST SAMPLE SHAPE HEATED TEMP. (°C) 

Structural information tests: 

FTIR flat piece (smooth surface) 150, 250, 500, 800 

UV-VIS-WIR flat piece (smooth surface) 150, 350, 500, 800 

BET powder (course ground) 200, 450, 750, 830, 860 

X-Ray powder (fine ground) 200, 450, 750, 800, 850 

Optical information test: 

Index of refraction polished flat piece 150, 450, 750, 800, 830 

Thermal information tests: 

DSC broken piece 150 

DTA broken piece 150 

TGA broken piece 150, 740 

TMA smooth cylinder's ends 150, 540 



Mechanical information tests: 



FLEX rectangular piece 150, 450, 750, 830 

COMP rectangular piece 150, 450, 750, 830 

DPN unpolished gel surface 150, 250, 450, 750, 800, 830 

Toughness unpolished gel surface 150, 250, 450, 750, 800, 830 

Density broken piece 150, 250, 450, 750, 800, 830 



76 

peaks in the sample's spectra. A dried gel was installed in a hot stage inside the FTIR 
sample chamber and heated to the temperatures designed in Table 3-1 for IR analysis. A 
heating rate of 3.3°C/min from room temperature to 800°C was used. 

The u!traviolet-visible-near infrared spectra were obtained from a Perkin-Elmer 
Lamda 9 UV/VIS/NIR spectrophotometer. This instrument consists of a high performance 
double-beam, double-monochromaior and a superior signal-to-noise energy optimized 
optical system [65] throughout the entire 185 to 3200 nm wavelength range; it is 
integrated with microcomputer electronics, video display, soft key operating system and 
printer. Gels heated to the temperatures designated in Table 3-1 and cooled to room 
temperature with heating programs shown in Figure 3-3 were taken out immediately 
from the furnace for testing. Subsequently, the thickness of the gels was measured and 
they were scanned at a rate of 120nm/min through a required wavelength range in 
either transmission or absorption mode after background correction had been made. 

The surface area, total pore volume, average pore radius, and pore size 
distribution were determined by the nitrogen adsorption-desorption isotherm BET 
method, using an automatic Quantachrome Autosorb-6 sorption system [66]. 

Specific surface area (A) of the gels is obtained from a series of data management 
and calculations performed in the microcomputer of the Autosorb-6 system. The 
calculations involve: (1) a BET equation, 1/{W[(P /P)-1 ] = 1/(W m C) + [(C- 
1)/(W m C)]x(P/P ) in which W is the weight of gas adsorbed at a relative pressure 
P/P (pressure ratio of N2 gas in He gas), Wm is the weight of adsorbate constituting a 
monolayer of N2 on surface, and the constant C is reiated to the energy of adsorption in 
the first layer. (2) a linear plot of 1/{W[(Po/P)-1]} vs P/P to yield values of slope 
s=(C-1)/(W m C) and intercept i=1/(W m C). (3) the weight of a monolayer W m obtained 
by equation W m =1/{s+i). (4) At=(W m NA cs )/M where At is total surface area of the 
sample measured and N is Avogadro's number. For N2 at 77 °K, the cross-sectional area, 



77 

Acs is 16.2 A 2 and M is the molecular weight of N2. (5) A=A t /W in which A is specific 
surface area of sample and W is the sample weight. 

The total pore volume (V|j q ) is derived from the amount of N2 adsorbed at a 
relative pressure close to unity, by assuming that the pores are all filled with liquidized 
N2 of a volume Vliq which can be calculated using equation (V|j q /V m )RT=P a V a d S where 
V m is the molar volume of the liquid N2, P a is ambient pressure, and V ac is is vaporized 
pore liquid (N2). 

The average pore size can be estimated from the pore volume, by assuming 
cylindrical pore geometry; then the average pore radius r p can be derived as r p = 
2V|j q /A. The pore size distribution is calculated using the method proposed by Barrett, 
Joyner and Halenda [67]. 

Samples heated to the temperatures designated in Table 3-1 and cooled to room 
temperature with the heating program shown in Figure 3-3 were ground into powder 
and weighed to around 0.6 gram in the pellet cells before installing in the Autosorb-6 
system for outgassing and preheating to eliminate the water moisture. The outgassing 
and preheating was held for 15 hours at 200°C in N2 gas atmosphere. Consequently, 
samples were transferred to the ports of the system for nitrogen adsorption-desorption 
isotherm measurements. Data were automatically accumulated in the mirocomputer and 
the results printed out . 

The X-ray diffraction analysis was obtained using a Philips diffractometer at room 
temperature with a 40Kv CuK a radiation and a nickel filter. The samples heated to the 
temperatures designed in Table 3-1 and cooled to room temperature with heating 
programs shown in Figure 3-3 were ground and scanned at a rate of 6°/min from 2e 
angles of 10° up to 50°. 

The index of refraction was obtained using a Pulfrich refractometer and a HeNe 
laser light source which wavelength is 632.8 nm. The principle of the refractometer is 
based on the measurement of the critical angle <t» c , which is the angle of the interface 



78 

between the unknown gel sample of index n and a prism of known index n'. Since n' is 
greater than n, the two must be interchanged in the standard equation, sin <t> c = n/n' 
[68]. The beam is oriented such that some of its rays just graze the surface as shown in 
Figure 3-4, so that the transmitted light has a sharp boundary occurs which allows one 
to compute the value of fa and hence of n. 

DSC, DTA, TGA, and TMA analyses were obtained with a DuPont 1090 thermal 
analysis system. In The DSC system, the gel sample and a reference were placed in pans 
which sat on a disk. Heat was transferred through the disk into the gel sample and 
reference. The differential heat flow to the sample and reference was monitored by the 
junction of a constantan disc and the chromel wafer which covers the underside of each 
platform. Chromel and alumel wires were connected to the underside of the chromel 
wafers, and the resultant wire-thermocouples were used to monitor the sample 
temperature. Therefore, heat transfer and temperature of the sample and reference 
could be recorded. The temperature range of the DSC cell is from room temperature to 
600°C. 

Differential thermal analysis (DTA) measures the temperatures at which heat- 
related phenomena occur in materials. DTA provides the same qualitative information as 
DSC, and can provide semiquantitative calorimetric measurements. The temperature 
range of the DTA cell is from ambient to 1200°C. 

The high temperature 1200°C DTA cell consist of a platinum sample and reference 
cups resting on the tops of two insulated thermocouple pedestals. The sample and 
reference were located 6 mm apart surrounded by a programmable furnace. 
Thermocouples located in the pedestals measured both the presence of transitions and the 
temperatures at which they occur. DTA cells complement the DSC to offer appropriate 
measurements over a wide temperature range. 

The thermogravimetric analyzer measures changes in weight as a function of 
temperature, and provides derivative TGA data. These data can be used to measure the 



79 



incident HeNe laser beam 
wavelength: 0.6328(im 



Sin O = n/n' 




Figure 3-4 Refraction in the prism of a Pulfrich refractometer. 



80 

changing in moisture and volatiles (oxidation reaction) when gel is in the heating 
process. 

A thermomechanical analyzer (TMA) can be used as a dilatometer to measure gel 
volume shrinkage, or glass expansion coefficient from room temperature to 800°C. The 
sample was installed in a programmable furnace in which a thermocouple in direct 
contact with the sample measured the sample temperature. A movable-core linear 
variable differential transformer (LVDT) whose output is proportional to the linear 
displacement of its core is used. The dimensional change of the sample with temperature 
can be monitored using this LVDT core displacement technique. 

Flexural strength tests were performed under guidelines of the ASTM D 790M-84 
standard [69]. Samples heated to the various temperatures (see Table 3-1) and cooled 
with the thermal schedule shown in Figure 3-3 were cut with a diamond watering blade 
and polished carefully with 600 SiC grit paper into a size of length x width x thickness 
(46 mm x 10 mm x 5 mm). All samples were dried at 150°C for 3 hours immediately 
prior to measurements to eliminate absorbed moisture. Subsequently, the samples with 
a span : width : thickness ratio of about 7:2:1 were loaded in three-point bending in 
ambient conditions at a strain rate of 3.5 x 10' 3 s -1 using an Instron model 1122. In 
this experiment a set of five identical samples were heated at same time in a furnace to 
each temperature. 

The compressive strength tests were carried out under the guidelines of the ASTM 
C1 58-80 standard [70]. Samples heated to the designated temperatures (see Table 3-1) 
and cooled with heating programs shown in Figure 3-3 were cut into a rectangular shape 
of length x width x thickness (14 mm x 7.5 mm x 5 mm). All samples were dried at 
150°C for 3 hours immediately prior to measurements to eliminate absorbed moisture. 
Subsequently, samples were loaded in an Instron model 1122 such that the length was 
parallel to the axis of the applied stress applied at a strain rate of 3 x 10" 4 s" 1 . The 



81 

same number of samples and processing temperatures were used as that of flexural 
strength test. 

Microhardness values were obtained using a 136° diamond pyramid indenter at a 
50 gram load with the Micro Hardness Tester, model M-400 F (Leco Co. Japan). 
Samples were heated to tSie designated temperatures (see Table 3-1) and cooled with 
heating programs shown in Figure 3-3. Then, the samples were placed under the 
indenter and applied with the 50 gram load. Two diagonals of the indenter were produced 
on the surface of the sample. The DPN can be calculated by measuring the average length 
of two diagonals through the microscope on the instrument. In this test five indentations 
were performed on each sample to obtain the data. 

Fracture toughness values were calculated using the extended crack lengths from 
the two stamped diagonals created by the diamond indenter on the surface of gel during 
the Vickers microhardness test. The calculations used to convert indentation length to 
fracture toughness are described by Anstis' relationship (Equation #16) [64] in the 
Results and Discussions of this chapter. 

Density of the samples was determined using a simple mercury displacement 
technique. Samples followed the heat treatments shown in Table 3-1 and Figure 3-3 
were immersed into a pycnometer. By knowing the sample weight, the corresponding 
weight of mercury displacement, and the density of mercury, the density of the sample 
was calculated. 

Results and Discussions 
Figure 3-5 shows the FT1R spectra for the partially densified gels heat-treated at 
various temperatures. The samples were scanned between 200 cm" 1 (50000 nm) and 
5600 cm" 1 (1786 nm). The results show that the Si-O-Si molecular stretching 



82 




5600 4400 3200 2000 1400 

Wavenumbers 



800 



200 



Figure 3-5 FTIR hot stage data from 25°C to 800°C of pure silica gel 



83 

vibration is observed at 1120 cm' 1 (8928.6 nm), even in the low temperature sample. 
The peak at 1250 cm- 1 (8000 nm) is an artifact of the diffuse reflection stage. The 
primary difference between these curves is that peaks corresponding to organic 
residuals in the range between 1400cm' 1 (7142.9 nm) and 2600 cm" 1 (3846.2 nm) 
are absent in the high temperature sample. The spectrum of the 800°C silica sample is 
nearly the same as that for fused silica, with the exception of a small shift in the 
absorption edge near 1400 cm' 1 (7142.9 nm) to tower wavenumbers. 

The temperature-dependent changes in intensity of the characteristic absorption 
band at 950 cm" 1 (10526.3 nm) have been attributed to the stretching vibration of the 
Si-O-H nonbridging oxygen (NBO) groups. With increasing temperatures, the 
concentration of siianol groups is decreased to a nondetectable level and the 
characteristic 950 cm' 1 (10526.3 nm) peak disappears. The extent of hydroxyl 
absorption bands at 3500 cm' 1 (2857.1 nm) to 4000 cm' 1 (2500 nm) is also 
diminished for the higher temperature samples. This does not mean that the gel is 
completely free (zero ppm) from all types of water, but rather that the FTIR technique 
is not sensitive enough in this region (950 cm' 1 ) to detect the residual hydroxyl bonds 
to fully understand and monitor the water associated with gel structure. Overtone and 
combination frequencies should be investigated [71]. 

These results show that the only significant "impurity" in the ultrapure silica gel 
is water. The amount of water determines the extent of non-bridging oxygen (NBO) 
content, which prevents complete densification. Water content can also be observed 
easily using a UV-VIS-NIR spectrophotometer. Figure 3-6 shows the intensity of free 
water peaks at 1363.3 nm, 1891.1 nm, and 2212.4 nm decreasing with increasing 
processing temperature, tt indicates that the densification is due to the combination of 
silanoi groups on the surface of particles which form free water and escape; 
consequently, the surface chemical water is reduced and the absorption peaks are 
diminished. 



84 



2.00 



1.60 



1.20 



CD 
O 

c 

CO 



9-0.80 



o 

JS 

C3 



0.40 



0.00 




200 



800 



1400 2000 

Wavelength (nm) 



2600 



3200 



Figure 3-6 The absorptance peaks of water decreasing with increasing temperature. 



85 

Samples heated to different temperatures are compared with a pure silica melt 
glass (Dynasil) in terms of the cut-off wavelength, as shown in Figure 3-7. Increasing 
the temperature of the thermal treatment increases the optical transmission near the UV 
absorption end and shifts the uv cut-off to the short wavelength for the pure silica gels, 
apparently as the result of a decreased water content in the high temperature samples. 

As Sigel concludes [72], the introduction of one electron valent elements (i.e. H, 
Li, Na, K, Rb, Cs, Fr, F, CI, Br, I) produces a noticeable shift of the uv edge to longer 
wavelengths. This shift is because these elements terminate the bridging oxygens (BO) 
into nonbridging oxygens (NBO) and provide lower energy exciton levels for photo- 
electron excitations. More water-related phenomena will be discussed in detail in the 
dehydration study in Chapter 4. 

Another important analytical technique for understanding the ultrastructure of 
partially densified gels is the N2 adsorption-desorption isotherm analyses. The results 
include analysis of the variation of average pore radius, pore radius distribution, 
specific surface area, and pore volume with densification temperature, as shown in 
Figures 3-8, 3-9, 3-10, and 3-11. There was no significant change in pore size 
(Figures 3-8 and 9) while the total pore volume and surface area decreased (Figures 3- 
10 and 3-11) with temperatures up to 860°C. An assumption is that the pores decrease 
in number and force the entire gel body to contract. This is because the pores are very 
small (in this study, the mean pore diameter is only 2.2 nm). Consequently, they 
essentially obey the mechanism presented by the Ostwald-Freundlich equation, 
log(S r /Sj) = KE/Tr, stated in Chapter 2 Equation #8 and illustrated in Figure 2-21. 
Once the pores start to decrease in size, the rate of decrease becomes very fast and they 
immediately fill and disappear under the assistance of the migration of silanol groups 
along the interior surface and/or migration of vacancies through the structure to the 
exterior of the gel. Therefore, the gel shrinks as the temperature increases as a result of 



86 



100 




Dash lines indicate the cut-off 
wavelengths 



200 



250 



300 350 

Wavelength (nm) 



400 



450 



Figure 3-7 Transmission cut-off of pure silica gel 



87 



OU " 






— — — ^^_____ _ 






^fl - 










£ 
















«< 

w- 40 - 














to 

3 
CO 

V— 

Q> 30 - 























o 

CL 












9n - 
















1 








i 


i ' h 


■• X I 




10-| 


J j UJ 


IP 1 J-fp-p 



200 



400 600 

Temperature (°C) 



800 



1000 



Figure 3-8 Pore radius vs. temperature 



88 




10 15 20 

Pore radius (A) 



25 



30 



Figure 3-9 pore size distribution vs. pore volume at various temperatures. 



89 



CO 

* 

i 

CD 

E 
§ 

CD 

| 

TO 
O 



0.40 



0.35- 



0.30 



0.25 



0.20 



0.15 



0.10 



0.05 



0.00 




200 



400 



600 



800 



1000 



Temperature (°C) 



Figure 3-10 Total pore volume versus temperature. 



90 



ra 
Si 



£ 
CO 

CD 

Q 

I 

3 

o 



600 



500 



400 



300 



1 200 

© 

Q. 

CO 



100 




200 



800 



400 600 

Temperature (°C) 
Figure 3-11 Specific surface area versus temperature. 



1000 



91 

the total pore volume decrease. It can also be reasonably assumed that the decrease of the 
surface area is linearly proportional to the disappearance of the number of pores. 

When a gel is heated higher than its foaming temperature, free water is formed 
from the dissociated surface hydroxyl groups inside the fully densified gel structure. 
Immediately, these free wafer molecules follow the idea gas law in Equation #1 to create 
new pores: 

pv=nRT (1) 

where p is internal pressure of a closed-pore volume v, n is a mole number of gaseous 
water molecules within an instantaneously created closed-pore v, v = 4rcr 3 /3 where r 
is the closed-pore radius, and T is gel body temperature at the moment foaming occurs. 
If N is the total molar number of gaseous water molecules in total of such created pores 
of V per unit volume of matter, then N/n is the total number of pores per unit volume of 
silica, and V = Nv/n is total pore volume (V VO jd) per unit volume of silica (V SO iid)- 
Consequently, equations #2 and #3 can be written: 

pV=NRT (2) 

V-(N/n) x v = (N/n) x 4rcr 3 /3 = (1- p)/p (3) 

where p, the relative density, is equal to p a / Pr , p a = m SO iid/(Vsolid+V V oJd) is the 
apparent density of the foamed silica gel and p r = m SO iid/V S olid is the fully densified 
silica gel. Therefore, from Equations (2) and (3), we get: 

p - 3nRT/4rcr 3 = NRTp/ (1- p) (4) 

When temperature exceeds the pore closing temperature, the gel immediately foams as 
soon as the surface water is released. 

The gel foaming mechanism is explored by J. Phalippou, T. Woignier, and J. 
Zarzycki {73]. They use the concept that the rate of total energy input to the gel 
sintering system equals the rate of total energy output from the system. The total energy 
input includes the surface energy of silica gel (dWa/dt = 8rcradr/dt) where r is the 
pore radius, or is surface tension, and t is time and the external pressure energy is 



92 

dWb/dt - -PdV = -P x 4jir 2 dr/dt, where P is external pressure. The total energy 
output includes the energy for viscous flow (dW c /dt = 16mirp(dr/dt) 2 ) where r\ is the 
viscosity of silica gel at the temperature of foaming, p is the relative density of the gel 
and the energy for varying the pore radius is dWd/dt = -pdV = -p x 4 7tr 2 dr/dt. The 
equation for this system is thus: 

dW a /dt + dW b /dt = dWc/dt + dW d /dt (5) 

By replacing all the items, we get: 

-2 a - r(P-p) = 4 rip(dr/dt) (6) 

and by combining with Equation #3 yields, 

2 a(1-p) 2/3 p 1/ 3 x (4Nn/3n) 1/ 3 + (P-p)(1- p)=4 Ti/3(dp/dt) (7) 
when we assume gel is sintered in conventional pressure, P=0, the equation becomes: 

dp/dt- (1- p)(3 a/2 nr - 3p/4 r\) (8) 

If there is no escape of gaseous water from the closed-pores, then combine Equation #4 

dp/dt- (1- p)(3 a/2 r\r) - 3 NRTp/4 ti (9) 

and let dp/dt= 0, and use Equation #3, then a critical pore radius r m j n is obtained: 

"min = (3nRT/8n a) 1/2 (10) 

By substituting Equation #10 into Equation #9, then, an expression for the maximum 
value of density (p ma x) is achieved: 

Pmax = 1/[(NRT/4 a )(3nRT/8rca) 1/2 + 1] (11) 

These two equations (#10,11) show that a maximum value p ma x and a corresponding 
critical pore radius r m j n can be predicted in terms of the sintering temperature (T), 
surface tension (a), the amount of free water in a pore (n) and the number of pores per 
unit volume of silica (N/n). From this study the conclusion is reached that whenever the 
residual surface water is released after the collapse or closing of the original open- 
pores, then the free water in the gel structure follows the idea gas law at higher 
temperatures to create closed-pores. Consequently, foaming of the gel happens and the 
average radii of the pores increases significantly when temperature is just above 860°C 



93 

(see Figure 3-9). At 860°C the pore radius suddenly increases from a 1.1 nm open- 
pore radius to a 5.4 nm closed-pore radius. 

X-ray diffraction patterns from fused silica generally exhibit a broad peak 
centered around the second strongest peak in the diffraction pattern of quartz (Figure 3- 
12). The partially densified silica gels made in this study have broader diffraction 
patterns than that of fused silica, as shown in Figure 3-12. The broadening of the gel 
diffraction peak decreases with increasing temperature, indicating an increase in the 
ordering inside the gel [74]. The BET data in Table 2-1 using Havard, Wilson, Iter's 
particle size model described in Section II also suggest that the effective particle 
diameter of the gels increases with temperature; e.g. 200°C (3.6 nm), 750°C (6.6 
nm), 800°C (7.1 nm) and 860°C (15.7 nm) [75]. These values can be compared to the 
diameter around 100 nm of fully densified silica. These results imply that very short- 
range-ordering is taken place inside the structure forming crystallites. The size of a 
single silica tetrahedron is about 0.3 nm. Therefore, the structure of the gel crystallites 
is composed of only few silica tetrahedra. The gel preheated to 200°C is estimated to be 
about 8 tetrahedra, at 750°C it is about 15 tetrahedra, at 800°C it is about 17 
tetrahedra, and at 860°C the gel has about 35 tetrahedra along the diameter of the gel 
fibrillar structure. As a result, x-ray diffraction produces a relatively broader peak 
for this relatively short-range-ordering than is observed for fused silica. 

This observation led to the suggestion that the silica gel is composed of a randomly 
oriented fibrillar structure (random-network model [23]) in which the silica 
molecules are relatively ordered crystallites (crystallite model [23]). This 
phenomenon is similar to a "mosaic structure" in an imperfect crystal in which the 
lattice is broken up into a number of tiny blocks (about 1000 A), each slightly 
disoriented one from another [74]. The overall observed gel structure is amorphous. 

Based upon the above results, the structure of porous gel in which the 
temperature-independent pore diameter is always around 2.2 nm (see Table 2-1) is 



94 



CO 
0) 




50 



40 



30 



20 



10 



Figure 3-12 X-ray patterns of silica gels at different temperatures 
compare to that of fused silica. 



95 

proposed as shown in Figure 3-13. The infrastructure of a densified gel is also proposed 
as shown in Figure 3-14. 

The data obtained show that the index of refraction of the gel monoliths increases 
with the pyrolysis temperature as well as density. The measured index of refraction 
ranged from n = 1.27 ± 0.03 to n =1.35 ± 0.04 for the sample heated from 150°C to 
830°C, and the corresponding density varied from 1.40 ± 0.02 g/cm 3 to 1.80 ± 0.02 
g/cm 3 . Within experimental error the results shown in Figure 3-15 are reasonably 
well predicted by the Lorentz-Lorenz equation [see p. 658 in ref. 23]: 

a=[3e (n 2 -1) M]/[No(n 2 +2)p] (12) 

Rearranging equation #12 yields: 

(n 1 2-i)/[(n 1 2 + 2 ) p1 ] = (n 2 2 -1)/[(n 2 2 + 2)p2] (13) 

where the constants are: 

a is polarizability of a silica molecule, 

e is the dielectric constant of a vacuum, 

M is the molecular weight of silica, 

N is Avogadro's number, 
and the variables are: 

n is index of refraction, p is density, 

ni is index of refraction of the gel and n 2 is that of fused silica, 

P1 is density of the gel and p 2 is that of fused silica. 
The results of Figure 3-15 show that it is possible that silica gel optics can be 
heated to specific temperatures to obtain a required combination of density and index of 
refraction. It is also possible that lenses can be obtained by controlling the temperature 
gradient in the silica gel to produce a refractive index gradient of refraction in a flat 
silica gel. 

The differential scanning calorimeter (DSC) data, shown in Figure 3-16, indicate 
an endothermic desorption of physical water at a maximum 100°C in the 27.7°C to 



98 



crystallite 
direction 




3.6 nm 
at 200°C 



Surface silanol groups are not shown 



Figure 3-13 A proposed gel structural model. 



97 



random orientation of 
relatively ordered crystallite 



20nm 



±6dr 




silanol group or nonbonding oxygen on the 
surface of relatively ordered crystallites 



Figure 3-14 A proposed scheme of sintered silica gel in which silanol groups 
terminate briding bonds on the surface of crystallites. 



1.6 



1.5 " 



g 1-4 



o 
x 



1.3 



1.2 



1 : Quartz 

2: Cristobalite 

3: Tridymite 

4: Dense gel-silica 
and fused silica 

5: 830°C gel-silica 

6: 750°C gel-silica 

7: 450°C gel-silica 

8: 200°C gel-silica 



98 




I 

2 



Density (g/cm 3 ) 



Figure 3-15 Index of refraction vs. density for silica gels, gel-glass, and 
crystallines phases. 



99 



12 



I 

Li- 
re 

CD 



-16 



- 


J3362.5°C 




27.7°C 






- \ / 




550°C 


Q^94.7°C 






_ .. L I 


I I 


I 



100 



200 300 400 

Temperature (°C) 



500 600 



Figure 3-16 The differential scanning calorimeter (DSC) data of a dried silica gel. 



100 

200°C range and an exothermic decomposition and/or oxidation of oxalic acid at a 
maximum 362.5°C in the 200°C to 450°C range. The DTA data are collected via the same 
method as the DSC data, but DTA has an increased testing range to 1200°C, as shown in 
Figure 3-17. There is significant endothermic desorption of physical water within the 
pores. A very large exothermic decomposition and/or oxidation of oxalic acid is also 
observed with DTA. No further thermal sorption is observed in the range between 550°C 
and 1200°C. Thus, the dried and desorbed gels are stable from 550°C onwards. 

A thermogravimetric analyzer (TGA) was also used to analyze the dried gel sample, 
as shown in Figure 3-18. In this case, the differential weight loss shows a very high 
peak at 124.0°C in the 25°C to 200°C range, indicating the maximum loss of physical 
water. There is a significant weight loss of oxalic acid at 361. 3°C in the 300°C to 400°C 
temperature range; no further weight loss was observed from 450°C to 800°C. This TGA 
observation, together with DSC or DTA data obtained during sintering, clearly indicates 
that two phenomena are present; (1) the endothermic water evaporation in the range of 
25°C to 200°C, and (2) the exothermic oxidation of oxalic acid in the 300°C to 400°C 
range. 

Figure 3-19 illustrates two TMA curves, one of an unfired standard silica gel 
sample and one of a fired (540°C) sample; the curve of the unfired sample has a 
significant decrease in linear dimension from 200°C to 750°C. The preheated 540°C 
sample shows only a slight dimensional decrease (0.056%) when reheated to 540°C. 
When heated above 540°C the dimensions decreases noticeably. These results show that 
the structure of the fired sample has already undergone rearrangement and that it is 
irreversible. 

In a 3-point bending test, a beam loaded has tensile stresses on one surface and 
compressive stress on the other, as shown in Figure 3-20. Flexural strength is a 
measure of the level of the tensile stress on the surface required to make a material fail. 
A partially densified gel is like fully densified glass which shows no plastic deformation 



101 



16 



12 - 



o4 



09 

O 

£ 
^0 

3 

|3 



£ 




200 400 600 800 

Temperature (°C) 



1000 1200 



Figure 3-17 The differential thermal analysis (DTA) data of a dried gel. 



102 



100 



90 



§80 

CD 



70 
65 




6.96% 
2nd wt. loss 



I 



1 



residues 
68.94% 



100 200 



300 400 500 

Temperature (°C) 



600 700 



800 



Figure 3-18 Thermogravimetric analysis (TGA) curve of a dried gel. 



103 



10 



-10 



7 

o 

- -20 
x 



a 



s -30 k 

rc 



2 
® -40 



£ 
b 



-50 



preheated 
540°C sample 




unfired sample 



1 



100 200 300 400 500 600 700 800 

Temperature (°C) 



Figure 3-19 Thermal mechanical analysis of a unfired sample and a preheated sample. 



104 



o 
3 

to 
,c 

(/) 
CO 

05 

S-. 
'5> 
0) 

"to 

c 
<u 




Compression 



Tension 



Figure 3-20 A three-point bending test. 



105 

under the stress. Samples can be loaded and stressed up to the proportional limit 
(rupture point at a maximum tensile stress a max ). Elastic strain is directly 
proportional to the applied tensile stress, a, by following the Hooke's law (a = E e where 
a is tensile stress, E is Young's modulus of elasticity for bending test, e is elastic strain. 
See ASTM D 790M - 84) and is recoverable below c max . In this test, the load to failure 
was calculated using the equation [see p. 156-158 in ref. 63]: 

c = 3P-[U2bd 2 0< a <a max (14) 

where 

a : tensile stress on the outer surface at midspan, pascal (newton/m 2 ), 

Pi : load at a given point on the load-deflection curve, N (newton), 

L : support span distance, m, 

b : width of specimen, m, and 

d ; thickness of specimen, m. 
the elastic strain before or at fracture was obtained using equation [69]: 

e= Zt = 6Rtd/L 2 = 6Dd/L 2 < e < e ma x (15) 

where 

e : strain, mm/mm, e max is the maximum strain at amax 

Z : strain rate, mm/sec, 

t : time, sec, 

R : rate of crosshead motion, mm/min, 

D : midspan deflection, mm. 
The variation of maximum flexural strength (a max ), maximum strain (e max ), 
Young's modulus of elasticity (E = a m ax/£max) with standard deviations for each set of 
five samples prepared at different temperatures are listed in Table 3-2 and shown in 
Figures 3-21, 3-22, 3-23. A large increase in flexural strength was noted above 
700°C (Table 3-2, Fig. 3-21). The specimen heated to 830°C has a value of about 30.4 
MPa which is about half the value of fused silica glass (58.9 MPa) [76]. The obtained 



106 



Table 3-2 
Mechanical properties of partially densified silica gels and fused silica 



Temperature 


Flexural strength 


Maximum strain 


Young's mc 


°C 


<*max ( Mp a) 


emax. (AL/L X10' 6 ) 


MPa 


150 


08.4 ± 0.5 


1060 ± 201 


07925 


250 


09.8 ± 0.7 


1071 ± 178 


09148 


450 


11.8 ± 1.2 


980 + 94 


12035 


750 


25.0 ± 2.3 


876 ± 157 


28539 


800 


27.0 ± 2.6 


833 ± 150 


32413 


830 


30.4 ± 3.6 


826 ± 121 


36803 



'"For reference [76]: 
fused silica 58.9 



806 



73089 



107 




200 400 600 800 

Temperature (°C) 



1000 



Figure 3-21 Flexural strength versus temperature. 



108 



1100 





1050 


<D 




LLI 




>< 




_1 




<1 


1000 






Cl> 




c 




3 








CI. 




3 






950 


p 








c 




w 




1— 








w 




F 


900 


3 




E 




X 




CO 





850 



800 




400 600 800 

Temperature (°C) 



1000 



Figure 3-22 Maximum strain to rupture versus temperature. 



109 



40000 




200 400 600 800 1000 

Temperature (°C) 



Figure 3-23 Young's modulus versus temperature. 



110 

maximum strain to the point of rupture for the gel samples decrease to approach the 
value, 806 x 10" 6 of vitreous silica (Table 3-2, Fig. 3-22). It is concluded that the 
gels have higher elastic deformability than fused silica since the fibrillar gel structure 
can endure relatively higher dimensional deformation before rupture. Because of the low 
densities of the porous gels, the Young's moduli calculated from equation E = a ma x/emax 
are much lower than the 73089 MPa [76] value of fused silica. However, the Young's 
modulus increases and approaches the value of fused silica as the densification 
temperature increases (Table 3-2, Fig. 3-23). We should also notice that it is very 
difficult to achieve a highly polished surface for partially densified gel-glasses. 
Consequently, relatively lower values with wider standard deviations are expected. 

The test results (Figure 3-24) show that the compressive strength increases 
gradually with temperature and reaches a value of 556 MPa at 830°C, approximately 
half the value of vitreous silica glass (1108 MPa) [76]. 

The density of the silica gels and gel-glasses are plotted as a function of firing 
temperature in Figures 3-25, with density increasing as the densification temperature 
increases. Only small changes in density were observed below 500°C; however, above 
700°C the density increased considerably with processing temperature. This indicates 
that viscous sintering is initiated above this temperature. The density of the samples 
heated to 830°C is about 1.80 ± 0.05 g/cm 3 - approximately 82% the density of fused 
silica glass. The temperature required to reach a density equivalent to type l-IV silica is 
a function of the ultrastructure of the gel itself, ranging from 830°C to 900°C, 
depending on particle size and the residual water content of the solid. 

The results of a diamond point microhardness test for silica gel as a function of 
pyrolysis temperatures are given in Table 3-3 and Figure 3-26. For a constant load 
(0.05 Kg), the length of the indentation diagonal decreases as the temperature and the 
microhardness increases. The 830°C gel sample has a microhardness value of 245 
Kg/mm2 which is about three times less than the 710 Kg/mm 2 value of fused silica. The 



111 



600 



500 



€ 400 

£ 

V) 
CD 
> 
CO 

to 

£ 

Q. 



300 - 



O 



200- 



100 




400 600 800 

Temperature (°C) 



1000 



Figure 3-24 Compressive strength versus temperature. 



112 



1.8 




1.3 



i I i i 1 — 

100 200 300 400 500 600 



— I 1 

700 800 900 



Temperature (°C) 



Figure 3-25 Density versus temperature. 



113 



Table 3-3 

Microhardness data of partially densified silica gel 

(0.05 Kg load) 



Temperature 


Indentation Diagonal length 


Microhardness 


CO 


d, (mm) 


DPN (kg/mm 2 


150 


0.0202 ± 0.0040 


113 ± 27 


250 


0.0193 ± 0.0054 


125 ± 38 


450 


0.0182 ± 0.0032 


140 ± 43 


750 


0.0154 ± 0.0024 


196 ± 52 


800 


0.0145 ± 0.0025 


220 ± 35 


830 


0.0138 ± 0.0030 


245 ± 49 



'Vickers' hardness number for silica is 710 kg/mm 2 [see p. 144 in ref. 63]. 



114 



300 



250 



CM 

* 

E 

E 
o> 

Z 
D_ 

Q 



200 



150 



100 




400 600 

Temperature (°C) 



1000 



Figure 3-26 Microhardness vs. temperature 



115 

slope of the curve in Figure 3-26 becomes very sharp at about 700°C indicating a 
significant structural change in gels. 

Fracture toughness indicates the amount of energy absorbed by a material during 
failure. This is in contrast to flexural strength (a ma x), which is a measure of the stress 
required to break a material. A tougher material can absorb more energy within the 
structure before rupture occurs. The critical stress intensity factor, K| C , can be 
estimated from the length of the plastic zone ahead of the crack tip. A simple testing 
procedure and economical method, using a Vickers diamond pyramid indenter, was 
introduced by S. Palmqvist [77] and evaluated by Anstis [64]. Although the 
determination of K| c by this method is not unique, the experimental relationship 
established by Anstis is given below: 

K| C = 0.016 (E/H)1/2 P 2 /(c/2) 3/2 (16) 

where 

E : elastic modulus in pascals, 

H: microhardness in Kg/cm 3 = 2P 2 sin 687d 2 [see p. 143-149 in ref. 63], 
P2 : indentation load in kg, 

68°: the half angle between opposite faces of the four-sided pyramid of 
diamond indender. 
d : diamond point indentation diagonal length, mm, 
c : extended crack length, urn. 
The experimental data (density, Young's modulus, microhardness and extended 
crack length) and the calculated results (Kj c and K| C /p) are compared to the values of 
vitreous silica, as listed in Table 3-4. The low K| c value (0.72 MPa-m 1/2 ) of fused 
silica glass indicates its brittle character. None of the silica gel samples has a higher K| C 
value than fused silica. This shows that the gel samples are even more brittle and easier 
to break than fused silica. Surpringly, the 150°C gel (0.49 MPa-m 1/2 ) is fairly 
tougher than the 830°C gel (0.40 MPa-m 1/2 ). If the K| C value is divided by the density 



116 



Table 3-4 

Fracture toughness (K| C ) and K| C /p ratio of partially densifiedsilica gels, data 

obtained from diamond indentation cracks. 

(0.05 Kg load) 



Temp. 

T 

(°C) 

150 
250 
450 
750 
800 
830 
fused silica 



Density 

P 
(*g/cm3) 

1.40 ± 0.05 
1.42 ± 0.05 
1.46 ± 0.05 
1.71 ± 0.05 
1.74 ± 0.05 
1.80 ± 0.05 
2.20 



Modulus 
E 
(MPa) 

7925 

9148 

12035 

28539 

32413 

36803 

73089 



Microhardness 

H 
(***Kg/cm2) 

11300 ± 2700 
12500 ± 3800 
14000 ± 4300 
19600 ± 5200 
22000 ± 3500 
24500 ± 9800 
71000 



Temperature 

T 

(°C) 


Extended Crack Length 
c 

** u.m 


Toug 
K| C 

*** i 


150 


24.3 ± 5.8 


0.49 


250 


27.0 ± 3.8 


0.43 


450 


30.6 ± 3.2 


0.39 


750 


32.2 ± 5.1 


0.47 


800 


35.0 ± 9.7 


0.42 


830 


35.9 ± 4.3 


0.40 


fused silica 


21.5 


0.72 



MPa-m 1/2 



K| C /P 

0.35 
0.30 
0.27 
0.27 
0.24 
0.22 
0.33 



* 1Kg = 1000 g. 

** 1 u.m = 10- 6 m * 10 _4 cm = 10- 3 mm, 

***1G.194 Kg/cm 2 = 1 MPa -145 psi. 



117 

of samples, the 150°C gel sample has a value 0.35 which is even greater than the value 
of 0.33 of fused silica. The reason is probably due to the fibrillar ultrastructure of the 
low temperature gel which is described in Figure 2-36, Chapter 2. 

Mechanical properties of the silica gels and gel-glasses, including flexural 
strength, compression strength, microhardness, and toughness, are all dependent on the 
gel ultrastructure. The evolution of ultrastructure, monitored by N2 adsorption- 
desorption isotherms, FTIR, uv-vis-nir, and x-ray diffraction techniques, proves that 
the pore volume, the surface area, and the amount of nonbridging oxygens (surface 
silanol group) decreases and the effective particle size increases with an increase in 
pyrolysis temperature. Consequently, the overall bulk density increases with sintering 
temperature, representative of the degree of ultrastructural rearrangement. In Figures 
3-27, 3-28, 3-29, 3-30, 3-31, and 3-32, the experimental data of compressive 
strength, maximum strain to failure, flexural strength, Young's modulus, 
microhardness, and toughness are plotted as a function of density. The maximum density 
of 2.2 g/cm 3 represents the value of vitreous silica. Within experimental error, the 
mechanical properties are linearly related to the density. A simple relationship is: 

Xgel = X s ( pge|/ps) (17) 

where 

Xgei : mechanical properties of the partially densified gel (i.e. compressive 
and flexural strength, Young's modulus, and microhardness), 

X s : mechanical properties of vitreous silica, 

Pgei : density of the partially densified gel, 

p s : density of vitreous silica, 2.2 g/cm 3 . 
Dashed lines drawn in Figure 3-27, 3-29, 3-30, and 3-31 were obtained from 
the above relationship. Although the present experimental results do not fit those 
predicted by equation 17, a linear relationship can still be applied to the present 



118 



1200 



Calculated value according to equation #17 



1000 



800 



CD 

a. 

^ 600 

"5s 



2 

■55 400 

CO 



200 




n •" 

1.4 



Experimental data 



1.6 



1 
1.8 



- T r 

2.0 



2.2 



Density (g/cc) 



Figure 3-27 Compressive strength versus density. 



119 



1200 



1100 __ 



1000 



to 
o 



§ 900 

o 



c 

E 

E 

I 
2 



800 



700 -J 




1.6 1.7 

Density (g/cc) 



1.8 



1.9 



2.0 



Figure 3-28 Maximum strength to failure versus density. 



120 



60 



50 _ 



40 _ 



re 

0- 



2 30 

CO 



2 



20 _ 



10 - 



-. 



1.2 



Calculated from equation #17 



\ 




A< 



Experimental data 



— ! 1 1 1 1 . 1 1 f 

1.4 1.6 1.8 2.0 2.2 



Density (g/cc) 



Figure 3-29 Flexural strength versus density. 



12! 



80000 



60000 - 



Q. 

Ǥ. 40000 

CO 

rj 

"5 
O 

E 

V) 

20000 



1.2 



..<< 



Calculated from equation #17 



1.4 



:♦• 




Experimental data 



i 

1.6 



i 

1.8 



— I— 
2.0 



2.2 



Density (g/cc) 



Figure 3-30 Young's modulus versus density. 



750 



650 - 



550 - 



122 



Calculated from equation #17 



%-' 



,<• 



CM 

* 

E 

I 



450 _ 



q 350 

CO 

CO 

CD 

c 

■s 

CO 



o 



250 



150 




Experimental data 



50 



1.2 



1.4 



1.6 



I 
1.8 



— — 
2.0 



-f 
2.2 



Density (g/cc) 



Figure 3-31 Microhardness versus density. 



123 



0.40 



0.35 



0.30 



I 



0.25 - 



0.20 _ 



0.15 




Density (g/cc) 



Figure 3-32 Toughness versus density. 



124 

experiment. 

Several models [see p. 773-777 in ref. 23] have been developed to predict the 
Young's modulus of a two phase system, such as a partially sintered gel-glass. The first 
is the Voigf model which assumes that the strain in each phase is the same; therefore, the 
Young's modulus of this two phase system is expressed as E up per bound = V2E2 + (1- 
V 2 )E2 where V2 is the volume fraction of the phase with modulus E2 , and E1 is the 
modulus of the other phase. The second is the Reuss model which assumes that the stress 
in each phase is the same; therefore, the modulus of this two phase system can be 
expressed as 1/E| 0W er bound = V2/E2 - (1-V 2 )/Ei. Z. Hashin and S. Shtrikman have 
established upper and lower limits for the moduli which are much narrower than the 
Voigt and the Reuss models. 

Ultimately, the second phase in a material can be considered as pore spaces that 
have zero Young's modulus value. This model was developed at porosities (closed pores) 
up to about 50% by J. K. Mackenzie and expressed as E/E = (1-1.9P + 0.9P 2 ) where P 
is porosity, and E is the modulus of the matrix phase. This is a much more reliable 
model compared to the first three models in dealing with porous material. Porous gel can 
be treated as a two-phase material in which the second phase is porosity. Consequently, 
it seems reasonable to use this model to predict the Young's modulus of the porous gel. 
The Young's modulus of the gels obtained from experiment are compared with the values 
from Mackenzie's model as shown in Figure 3-33. 

The experimental values of Young's moduli for the gels are generally lower than the 
predicted values. The deviations between them at lower densities are larger than that at 
higher densities. This may be due to the high surface water content in the lower density 
gels which promotes crack propagation [24]. As the gel becomes denser, the number of 
pores decreases and the surface water is reduced. Therefore, in the higher density 
region, the experimental data become closer to the values the above equation predicts. 



125 



1.0 



0.8 - 



0.6 - 



CO 

jg 

ZJ 
t3 
O 



CO 

£2 

<B 
W 

O) 

C 

O 
> 

re 
c 
o 



0.4 



§ 0.2 



0.0 -. 



MacKenzie's model 



0.0 




r 

0.2 



Volume fraction pores 



Figure 3-33 Relative Young's modulus versus porosity. 



126 

Conclusions 

The determination of the physical properties of partially densified gels establishes 
the nature of the porous gel ultrastructure and ultrastructural dependence of properties. 

FTIR analysis showed a 950 cm- 1 SiOH stretching vibration peak decreasing with 
increasing temperature, indicating that the sample is becoming increasingly dehydrated. 
The peaks of organic residuals in the range from 2000 cnr 1 to 3000 cm" 1 disappear as 
temperature increases. The shift to lower UV cut-off wavelengths with increasing 
temperature, noted in the UV-VIS-NIR data, also shows that the impurity (water) level 
is reduced. A quantitative study on the change of water level during sintering is discussed 
in Chapter 4. 

X-ray diffraction of the gels showed no evidence of devitrification, confirming the 
development of an amorphous glass phase from the gel. More important in this study, the 
observation led to suggestion that silica gel is composed of random oriented fibrillar 
structure (random-network model) in which the silica molecules are very well ordered 
crystallites (crystallite model). 

The index of refraction of silica gel varied with density as predicted by the 
Lorentz-Lorenz relationship. This variation of refractive index with density can be 
utilized to manufacture optical waveguides and lenses using localized index gradients. 

BET data showed a uniform size distribution of porosity for all temperatures below 
the fully densified temperature. The densification mechanism reduces the volume and 
number of pores, as opposed to pores merging together without reduction in pore 
volume. 

The DSC, DTA, TGA, and TMA data provide information useful in monitoring 
thermodynamic, weight loss, and dimensional changes during sintering. With these data 
an optimized manufacture process can be achieved. 

Mechanical properties, including flexural and compressive strength, 
microhardness, together with density, showed an increase in values with increasing 



127 

processing temperature approaching the values of vitreous silica. The low toughness, K| C 
and Kjc/p, values of the partially densified gels are comparable to those of Type l-IV 
vitreous silicas. The interesting point is the 150°C gel sample has a higher Ky P value 
than fused silica confirming that the fibrillar ultrastrucfure of the gel can absorb 
relatively high energy before rupture occur. The presence of surface water is suggested 
to be a major deteriorating factor for the mechanical properties, and is especially 
severe in the lower temperature gels. 



CHAPTER 4 
DEHYDRATION OF SOL-GEL DERIVED SILICA OPTICS 



Introduction 
An amorphous silica gel can be characterized by a random packing of Si02 
tetrahedra which gives rise to a nonperiodic, solid, fibrillar structure with many voids 
and a very large surface area. The surface area ranges from 500 m 2 /g to 900 m 2 /g 
depending on the low temperature sol-gel processing schedule. 

Based on ller's study, silica gel consists of connected spherical particles; the 
interior of the particles have a density of 2.2 g/cm 3 made entirely of anhydrous -O-Si- 
O-Si- bridging bonds. Located on the particle's surface are non-bridging terminal 
oxygens, each having an attached hydroxyl ion; these are also referred to as silanol 
groups. As a wet silica gel is dried, or partially densified, free water is removed from 
the pores; however, the siianoi groups remain intact [78, 79]. 

Theoretically, for ultrapure silica without silanol groups the energy gap between 
the valence and conduction bands is approximately 8.9 eV, as the oxygen ions have very 
tightly bound electrons [80]. The high intrinsic absorption edge results from the 
excitation of the valence band electrons within the bridging Si-0 network to unoccupied 
higher energy states, such as exciton levels or conduction band levels [see p. 161-164 
in ref. 26]. To excite these electrons requires ultraviolet photons of at least 140 nm 
wavelength (or wavenumber - 71428 cm" 1 ). Thus, the UV absorption peak for 
ultrapure silica should occur at approximately 140 nm and its UV absorption tail, which 
is associated with thermally activated phonons [81], becomes negligible in the visible- 
infrared portion of the spectrum. 



128 



129 



The intrinsic fundamental vibrations of the ultrapure silica molecules result in 
resonance with the incoming light at an infrared absorption peak of 8333 nm (1200 
cm" 1 , 0.149 eV); however, weak combination and overtone bands exist at 3200 nm 
(3125 cm- 1 , 0.39 eV) and 3800 nm (2632 cm" 1 , 0.33 eV), and strong bands occur at 
4400 nm (2273 cm" 1 , 0.28 eV). The infrared absorption tail of ultrapure silica, like 
its UV absorption tail, is also caused by phonons. These combinations and overtones 
influence the infrared absorption tail down to 1300 nm (7692 cm- 1 , 0.95 eV) [82]. 

Extrinsic absorption of light in silica gel in the 140 nm to 8333 nm range has 
been detected and interpreted as essentially the result of surface hydroxyl groups and 
their associated free water; only one ppm of hydroxyl ions in glass can produce 30 
dB/km loss at 1390 nm [83]. All other types of impurities have been reduced to very 
low levels (only several parts per billion) by a chemical refining system during TMOS 
synthesis, thereby contributing no significant absorption effects in these gels. 

Thus, a major problem in producing gel-silica optics is that gel surface hydroxyl 
groups and hydrogen-bonded pore water give rise to atomic vibrational energy 
absorption in almost the entire range of ultraviolet to infrared wavelengths (160 nm to 
4500 nm). This absorption greatly decreases the optical applications of a silica-gel 
monolith. Consequently, in order to achieve the full theoretical performance of silica 
complete dehydration is imperative. The degree of dehydration of gel-silica optics is 
monitored by analyzing the light absorption spectra in a broad range; the Perkin-Elmer 
UV-VIS-NIR spectrophotometer covers the range from 184.5 nm (54200 cm' 1 ) to 
3200 nm (3125 cm" 1 ) and the Nicolet FTIR covers the range from 2083 nm (4800 
cm' 1 ) to 50,000 nm (200 cm' 1 ). 

After extensive experimentation a reliable method was found that completely 
eliminates the surface chemical hydroxyl groups and associated pore water in gel-silica 
monoliths. By applying the concepts of fundamental silica surface chemistry [84-86], 
it was found that many chlorine compounds - some of these include methylated 



130 

chlorosilanes, such as CISi(CH 3 ) 3 , Cl2Si(CH 3 ) 2 , CI 3 Si(CH 3 ), silica tetrachloride 
(SiCU), chlorine (CI2) and carbon tetrachloride (CICl4)-can completely react with 
surface hydroxyl groups to form hydrochloric acid, which then desorbs from the gel 
body at a temperature range (400°C to 800°C) where the pores are still interconnected. 
In this study, carbon tetrachloride is used successfully to achieve complete dehydration 
of ultrapure gel-silica monoliths. 

Review of the Literature Re garding Dehydration 
The quality of silica gel can be significantly reduced by impurities. By far the most 
troublesome impurity, "water", is present in two forms: free water within the 
ultraporous gel structure (i.e., physical water), and hydroxyl groups associated with 
the gel surface (i.e., chemical water). The amount of physical water adsorbed to the 
silica particles is directly related to the number of hydroxyl groups existing on the 
surface of silica. During the 1950's and 1960's, researchers Young, Fripiat, Benesi and 
Jones, Hockey and Pethica, Kiselev, McDonald, et al. [87-91] contributed much 
information regarding the hydration/dehydration characteristics of the silica gel/water 
system, as summarized below: 

1. The physical water can be eliminated and surface silanol (Si-O-H) groups 
condensed starting at about 170°C, as shown in Figure 4-1. Thermal analyses, such as 
TGA and DSC, confirmed this process in our silica gel system, as shown in Chapter 3. 

2. The dehydration is completely reversible, up to about 400°C, as shown in 
Figure 4-2. Decomposition of organic residuals, up to 400°C, was also confirmed using 
DSC and TGA for our TMOS derived silica gels, as presented in Chapter 3. 

3. Above 400°C, the dehydration process is irreversible as a result of 
shrinkage and sintering across pores, as shown in Figure 4-3. Thus, the amount of 
existing hydroxyl groups on the gel surface is an inverse function of the temperature of 



131 




(25°C < > 1 70°C) 



+ n H 2 



Figure 4-1 Physical water decreases and silanol groups condense 
in the range of room temperature and 170°C. 



132 



Reversible 



(1 70°C < > 400°C) 



2698.9 nm 



\) 2 =2732.24 nm 




2768.9 nm 



-0.= 2668.80 nm 



+ n H 2 



Figure 4-2 Surface silanol groups are reversible in the range of 170°C to 400°C. 



133 




Figure 4-3 Irreversible elimination of adjacent hydroxyl groups. 



134 

densification. It is shown in Chapter 3, based upon UV-VIS-NIR data, that the reduction 
of surface hydroxyl groups occurs above 400°C. 

4. Viscous flow occurs above 850°C with the exact temperature depending on the 
particle size of a specific gel. The singular hydroxyl groups on the gel surface react with 
each other bringing particles together, thereby eliminating voids within the gel. Some 
surface water, which is unable to be desorbed prior to pore closure, is trapped inside 
the densified gel. 

Young, in his early work, found that the decrease in surface area of the silica gel at 
high temperatures is a function of the time and temperature of the heat treatment. This 
supports the concept that the sintering mechanism is essentially the result of viscous 
flow, rather than surface diffusion. Impurities (i.e., surface water) effectively lower 
surface energy and thereby the sintering temperature, presumably by facilitating 
viscous flow; Zarzycki, et al. [73] confirmed this point. 

Hair [see p. 87 in ref. 27] also proved that heating silica gel in the 170°C to 
400°C range causes reversible dehydration via elimination of surface water and the 
formation of both single and adjacent surface hydroxyl groups, as illustrated in Figure 
4-2. Hair found that at 400°C, no more than half of the surface hydroxyl groups had 
been desorbed and that most of the remaining surface hydroxyl groups were adjacent to 
each other and therefore situated for preferential water adsorption, shown in Figure 4- 
4. He stated that heating the gel above 400°C causes a drastic, irreversible elimination 
of adjacent hydroxyl groups, as shown in Figure 4-3, until at about 800°C, only single 
hydroxyl groups remain, as shown in Figure 4-5. As the temperature increases, single 
hydroxy! groups depart from the gel surface until the gel is densified; this occurs in the 
850°C to 1000°C range. However, some single hydroxyl groups are still unable to 
escape from the gel surface and therefore can contribute to foaming of the gel as the 
temperature increases. 



135 



Cool below 400°C 




+ n H 2 




"~<Z/ 



d 4 =2919.20 nm 



u 3 =281 6.88 nm 



Figure 4-4 Reabsorption of physical water below 400°C. 



136 




"Ui= 2668.80 nm 




Figure 4-5 Only single hydroxyl groups remain at temperature above 800°C 



137 

More importantly, Hair mentioned that when the silica gel has been completely 
dehydrated, there are no surface hydroxyl groups to adsorb the free water; in other 
words, the surface is essentially hydrophobic. Clearly, it is the realization of this 
critical point that is the focus for this study. 

The vibrational overtones and combinations of hydroxyl groups and their associated 
molecular water, occurring in the 1250 nm to 3000 nm range, have been studied by 
Anderson and Wickersheim [92]. Evaluation of a partially dehydrated (800°C) silica gel 
shows an absorption peak at vi - 2668.80 nm (see Fig. 4-5), surely due to the 
fundamental stretching vibration of hydroxyl groups on the gel surface. These singular, 
or free, hydroxyl groups are also referred to as "isolated silanol groups". The 
symmetrica! appearance of this peak indicates that these singular hydroxyl groups have 
no interaction with water molecules. The band at 1366.12 nm (2u2) is the first 
overtone of the adjacent silanol group vibration x>2 - 2732.24 nm (see Fig. 4-2). The 
1366.12 nm peak becomes less intense as the gel is heated and disappears with complete 
dehydration. The combination peak at 2207.51 nm (v2 + \>0H (bend)) is the result of 
the hydroxyl ion's stretching and bending vibrations, where ■uoh is a bending wavelength 
between 11494.25 nm and 12345.67 nm. Researcher Peri [93] suggests that this 
combination band is due to the Si-O-H stretching vibration and an out-of-plane O-H 
displacement (bending) vibration. This type of hydroxyl group is labeled an OH(2) 
group. 

The adjacent hydroxyl groups also interact with free water (V3, see Fig. 4-4) to 
form hydrogen bonds; this effect causes a change in both the fundamental stretching 
vibration and its associated overtones and combinations. Therefore, the hydroxyl groups 
associated with water show a new combination peak at 2262.44 nm (\>3 + uoh (bend)); 
this kind of hydroxy! group is called QH(3). The energy calculations, by Benesi and 
Jones [88], predict that the fundamental stretching vibration of OH(3) at 1)3 = 2816.88 
nm is a value shifted about 148.08 nm from the vibration of the free hydroxyl group at 



138 

ui = 2668.80 nm (OH(1)). From actual absorption data, McDonald observed a peak at 
2816.88 nm, indicating a strong interaction between free pore water and surface 
hydroxyl groups. 

When a dehydrated silica gel is exposed to a slightly humid air atmosphere, sharp 
peaks appear at 2816.88 nm (t) 3 ), 2732.24 nm (\) 2 ), 1890.35 nm (t> 3 + 2voh 
(bend)), 1459.85 nm (2104), and 1408.44 nm (2\>3). Hair [see p. 89 in ref. 27] 
believes that the intensity changes upon adsorption of water indicate that all these bands 
are connected with the hydroxyl group which is associated with physical pore water. 
Further hydration results in a broadened band at about x>4 = 2919.70 nm (see Fig. 4-4) 
--characteristic of bulk water. 

Cant and Little [94, 95], and Chapman and Hair [96], tend to agree that for silica 
gel a sharp and slightly asymmetrical peak on the high-wavelength side, at 2668.80 nm 
(ui), together with a distinct band at 2732.24 nm (x>2), can be attributed to freely 
vibrating surface silanol groups, and to hydrogen-bonded adjacent silanol groups, 
respectively. In addition, a broad band at 2919.70 nm (^4) is due to the stretching of 
molecular water. 

Elmer, et al. [97], in their rehydrated study of porous silica showed that the 
intensity of the peak at 2668.80 nm increases during rehydration. They also indicated 
that physical water prefers to adsorb on adjacent hydroxyl groups rather than on the 
singular hydroxyl groups. 

Recent studies in optical fiber communication technology by D. B. Keck, R. D. 
Maurer, and P. C. Schultz [71] found that the extrinsic hydroxyl groups also give rise 
to some noticeable overtones and combinations occurring roughly at 725 nm, 880 nm, 
950 nm, 1125 nm, 1230 nm, 1370 nm. These absorptions strongly degrade the 
performance of optical fibers. 

Most of the silica glasses manufactured by melt or synthetic methods (Type I to IV 
silicas stated in Chapter 1), such as those produced by Corning, Melles Groit, Dynasil, 



139 

and Quartz Scientific, Inc., result in impurities (e.g., water and/or metallic elements). 
Three significant absorption peaks at 2732.24 nm (^2), 2207.51 nm (m 2 + 
DOH(bertd)} and 1366.12 nm (2u 2 ) are found to be the unique stretching vibration of 
adjacent silanol groups and its overtone and combination, as shown in Figures 4-6, 4-7, 
4-8, and 4-9. No singular silanol group (ui) was found using high resolution UV-VIS- 
NIR spectrophotometer. 

The electrons of these impurity atoms can be easily excited by photons of lower 
energy than those associated with the 8.9 eV UV band edge of theoretically pure silica, 
thereby causing a shift in the ultraviolet absorption edge to longer wavelengths. These 
excitations also cause additional absorption bands or peaks in the visible and near 
infrared ranges. Without complete dehydration, the quality of silica gel-glasses is 
significantly affected by the problem of water retention. 

The highest quality of pure silica manufactured in the world today is that of optical 
fibers fabricated by vapor phase reaction of pure oxygen with silicon tetrachloride 
(Type IV silica). This process results in fibers of ultralow loss - about 1 .0 dB/km to 
5.0 dB/km in the 900 nm to 1300 nm range. It is shown in Chapter 5 that the fully 
dehydrated, completely densified, gel-glass monoliths developed in this dissertation are 
of such a quality as to compare with optical fibers. 

Experimental Procedure 
The standard dried gels (150°C), manufactured as per Example One in Chapter 2, 
are used as the basis for preparing two sample sets for the following dehydration study. 
One sample set was partially densified at designated temperatures in an ambient air 
atmosphere; the other set was chemically and thermally treated prior to sintering in a 
mixed vapor (carbon tetrachloride and helium) atmosphere within a special apparatus, 
shown in Figure 4-10. 



140 



100 



C 

g 

to 

E 
to 
c 
o 

H 




320 400 1000 

Wavelength nm 



3000 



5000 



Figure 4-6 Transmission curve from Corning Glass Co. commercial UV grage optical 
melt silica Code 7940. Thickness 10 mm. 



141 




320 400 1000 

Wavelength nm 



3000 



5000 



If 



Figure 4-7 Transmission curve from Melles Griot Co. commercial UV grage optical 
melt silica Code UVGSFS. Thickness 10 mm. 



142 




320 400 1000 

Wavelength nm 



3000 



5000 



Figure 4-8 Transmission curve from Dynasil Co. commercial UV grage optical 
melt silica Code 1000. Thickness 10 mm. 



143 




240 



320 400 1000 

Wavelength nm 



3000 



5000 



Figure 4-9 Transmission curve from Quartz Science Inc. commercial UV grage optical 
melt silica. Thickness 10 mm. 



144 



A 



Sample 




Exhaust 



Thermocouple 




Helium 



Figure 4-10 A mixed vapor(CCl4 and He) atmosphere within the tubing of a furnace. 



145 

Densification in an air atmosphere was carried out using the heating program 
shown in Figure 4-11. These samples were heated to designated temperatures (150°C, 
450°C, 750°C, 800°C, 850°C), cooled to room temperature, and then subjected to 
density and optical absorption measurements. The UV-VIS-NIR and FTIR spectra were 
used to monitor the fundamental, overtone, and combination vibrations of hydroxyl 
groups within the ranges of 900 nm to 3200 nm and 2083 nm to 50,000 nm, 
respectively. 

The heating program for densification of samples dehydrated in a carbon 
tetrachloride/helium atmosphere in a tube furnace is shown in Figure 4-12. During the 
dehydration process carbon tetrachloride was consumed at a rate of 4 cc/hour. The 
samples were removed at various temperatures during the heating program (850°C, 
950°C, 1050°C, 1150°C), Density measurements of the samples were taken, followed 
by the UV-VIS-NIR and FTIR spectra measurements within the previously stated ranges. 

Results and Discussions 

The density measurements at various sintering temperatures for samples with or 
without chlorination are shown in Figure 4-13, in which the density of the water-rich 
(without chiorination) gel sample reaches a maximum (= 2.2 g/cc) at a lower 
temperature about 860°C, and the density of the water-free (with chlorination) gel 
sample has its maximum (= 2.2 g/cc) at a relatively higher temperature of about 
1100°C. This indicates that the hydroxyl groups significantly decrease the sintering 
temperature by lowering the surface energy of silica. 

The important absorption peaks and bands found in this dehydration study are 
summarized in Table 4-1. These peaks and bands are identical to those discovered by 
previous researchers stated in Section II of this Chapter. 

Curves a, b, c, and d in Figure 4-14 show the UV-VIS-NIR spectra of gels heated in 
ambient air at various temperatures up to about 850°C. Overtone and combination 



146 



1000 



800 



600 



O 

3 400 



CD 
Q. 

E 

CD 

H 



200 




120 



Time (hour) 



Figure 4-1 1 Heating cycles for air atmosphere furnace. 



147 



1200 




Time (hour) 



Figure 4-12 Four heating programs for controlled atmosphere furnace. 



2.4 



148 



2.2 



2.0 



o 
o 

D) 1.8 



"55 



CD 

Q 



1.6 



1.4 



1.2 



samples densified in air 
atmosphere 




samples densified in CCI 4 

atmosphere 



J i_^ 



J i_i 



! 



200 400 600 800 

Temperature (°C) 



1000 1200 



Figure 4-13 Density measurements at various temperatures for samples with or 
without CCI4 treatment. 



149 



Table 4-1 
Absorption peaks of the pore water and the surface hydroxyl groups of gel-silica 

monoliths 



Wavelength Identification observation command 

(nm) 

2919.70 ••**•„ 4 a broad peak on a broad band 

2816.88 ****\>3 a tiny peak on a broad band 

2732.24 ***\)2 a joint of two small peaks at 

2768.90 nm and 2698.90 nm 

2668.80 **i>i a very sharp symmetric peak 

2262.48 x>3 + *\>oh abroad band, no peak 

2207.51 \>2 + 'uoh a high broad asymmetric peak 

1 890.35 V3 + 2uoh a high broad asymmetric peak 

1459.85 2\>4 a tiny peak on a broad band 

1408.44 2\>3 a small peak on a broad band 

1366.12 2^2 a very sharp symmetric peak 

1237.85 {[2\>3 + •uoh] + a small peak 

[2l»2 + V0H]}/2 

1131.21 2u3 + 2voh a tiny peak 

938.95 3u 3 a small peak 

843.88 3\>3 + doh no peak observed 

704.22 4\>3 a tiny peak 

*t»OH : an out of plane bending vibration of Si-O-H bond. 

**x>-\ : stretching vibration of an isolated Si-O-H bond. 
** \>2 : stretching vibration of an adjacent Si-O-H bond. 

"v3 : stretching vibration of a Si-O-H bond which is hydrogen-bonded to water. 
"■U4 : stretching vibration of absorbed water. 



150 



2.00 




20C 



800 



1400 2000 

Wavelength (nm) 



2600 



3200 



curve a is the spectrum of 150°C sample 
curve b is the spectrum of 750°C sample 
curve c is the spectrum of 800°C sample 
curve d is the spectrum of 850°C sample 



Figure 4-14 Absorption curves of partially densified gels in air. 



151 

vibrational peaks are observed at 704.22 nm, 938.95 nm, 1131.21 nm, 1237.85 nm, 
1366.12 nm, 1408.44 nm, 1459.85 nm, 1890.35 nm, 2207.51 nm. A very strong, 
broad absorption band occurs between 2400 nm and 3200 nm. None of these peaks have 
been eliminated by heating, instead they have only decreased in intensity with increasing 
temperatures. Clearly, the gel is not completely dehydrated, even when heated to the 
point of full densification; further heating results in a foaming problem. 

Data obtained in this study show that a combination vibration is identified at 
2207.5 nm, resulting from the adjacent silanol stretching vibration at 2732.24 nm 
(x>2) and the out-of-plane hydroxyl ion deformation vibration at 11494.25 nm (uoh 
(bend)). The peak at 1890.35 nm is a combination vibration of 2816.88 nm (\>3) plus 
two times the bending frequency (2uoh (bend)). The peak at 1459.85 nm (2va) seems 
to be the first overtone of the 2919.70 nm (1^4). The peak at 1408.44 nm (2t>3) 
observed is the first overtone at 2816.88 nm (t»3); whereas the 1366.12 nm (2\>2) 
peak is exactly from the first overtone of the fundamental hydroxyl stretching vibration 
observed at 2732.24 nm (^2)- 

The peak observed at 1237.85 nm is presumed to be an overlap from the 
contribution of two type of combinations which are 1221.00 nm (2^2 + vqh (bend)) 
and 1254.70 nm (2\>3 + doh (bend)). A tiny peak at 1131.21 nm is believed to be 2u 3 
+ 2i>oh (bend) and a small peak at 938.95 nm is presumed to be a second overtone of 
2816.88 nm (3t>3). There is a very tiny peak at 704.22 nm which is a third overtone 
of 2816.88 nm (4\)3) as shown in Figure 4-14 curve d. 

These results show that for critical optical applications where complete 
transmission over a broad range of wavelength is important, densification in an air 
atmosphere is obviously a failure. The resulting quality of this gel can not compete with 
that of fused silica (see Chapter 1), and it will never reach the point of complete 
dehydration. 



152 

Carbon tetrachloride treated samples were removed from the tube furnace after 
reaching various temperatures (850°C, 950°C, 1050°C, 1150°C) and then analyzed to 
determine their characteristic UV-VIS-NIR absorption spectra, as shown in Figures 4- 
15, 4-16, 4-1 7(a) and (d). Absorption peaks were visible at 2890.1 nm, 2768.9 nm, 
2698.9 nm, 2668.8 nm, 2207.5 nm, 1897.6 nm for the 850°C sample; and at 2884.3 
nm, 2765.4 nm, 2698.3 nm, 2669.4 nm, 2207.5 nm, 1897.6 nm for the 950°C 
sample. 

Stretching vibrations of the adsorbed physical water gives rise to typical broad 
absorption peaks at 2890.1 nm and 2884.3nm which are shifted from 2919.70 nm (u 4 ) 
within a broad range from 2700 nm to 3200 nm. Absorption peaks at 2698.3 nm and 
2698.9 nm are suggested to be the result of the stretching vibrations of hydrogen- 
oxygen bonds of adjacent silanol groups. The 2768.9 nm and 2765.4 nm peaks are 
proposed to be the result of stretching of the hygrogen bonds to the neighboring silanol 
oxygens, as shown in Figure 4-2. These two kinds of absorption peaks in general can not 
be distinguished and thus form the combined broad peak at 2732.24 nm which is 
observed by many researchers [91, 93, 98, 99]. The sharp peaks at 2668.8 nm and 
2669.4 nm are identified to be caused by vibrating surface isolated silanol groups (i.e., 
free hydroxyl groups). 

The intensity of all absorption peaks decreases as the temperature increases. The 
spectrum from the 1050°C sample shows only one peak, as shown in Figure 4-17 (a), 
occurring at 2668.8 nm {v\), which is caused by isolated hydroxyl groups. The sample 
heated to 1150°C has a spectrum in which the water peaks have been eliminated, as 
shown in Figures 4-17 (b) and 4-18. The absorption loss due to water is estimated to 
approach zero as no water or hydroxyl absorption peaks are present at any wavelength. 
The quality of optical transmittance of this sample is significantly higher than that of 
traditional fused silica glass. 



153 



1.00 



0.80 



0,0.60 

o 

c 

-»— « 

§"0.40 

CO 

3 



0.20 



o.oo I l 



200 



800 



2207.5 nm 
1897.6 nm 




1400 2000 

Wavelength (nm) 



2600 3200 



Figure 4-15 Absorption curve of gel partically densified in controlled CCI 4 atmosphere for 
a 850°C sample of 4 mm thickness. 



154 



1.00 



0.80 



0.60 

8 

c 

eo.40 

o 

_Q 



0.20 



0.00 



2669.4 nm- 
2816.9 nm — 



200 800 



2698.3 nm 
2765.4nm 



2884.3 nm 

2207.5 nm 

1897.6 nm . 



jl. 



j_ 




j L 



1400 2000 2600 3200 

Wavelength (nm) 



Figure 4-16 Absorption curve of gel partically densified in controlled CCI 4 atmosphere 
for a 950°C sample of 3.8 mm thickness. 



155 



0.40 
0.20 



(a) 1 050°C sample 



2668.8 nm 



V__ 



o 0.00 
c 200 

CO 



I I 



800 1400 2000 

Wavelength (nm) 




2600 



200 800 1400 2000 
Wavelength (nm) 



2600 



3200 




3200 



Figure 4-1 7 Absorption curves of gels particaHy densified in controlled CCI 4 

atmosphere for a 1050°C sample of 3.6 mm thickness and 
a 1 1 50°C sample of 3.4 mm thickness. 



156 




4000 3500 



2500 2000 1500 1000 500 



Wave numbers (cm" 1 ) 



Figure 4-18 FTIR absorption curve of fully densified gel-glass. 



157 

Samples which had been heated to 1050°C in the tube furnace were aged for 
various durations in air: 1 day, 2 days, 4 days, and 7 days. The density, surface area, 
total pore volume, and pore radius measured were 1.89 g/cm 3 , 187.23 m 2 /g, 0.14%, 
and 11.07 A respectively. The resulting absorption spectra from each of these samples 
indicates the readsorption of molecular water with the corresponding reappearance of a 
broad absorption peak in the 2863.2 nm to 2898.5 nm range and shows no overtone or 
combination peak, as shown in Figure 4-1 9(a), (b), (c), and (d). 

On the other hand, the samples which were heated to 1150°C in the tube furnace 
and aged in air for 7 days, 14 days, and 30 days showed no evidence of readsorption , as 
shown in Figure 4-20(a), (b), and (c). Consequently, the dehydration and densification 
of gel-silica monoliths as developed in this study results in an optical material 
equivalent to the best Type IV silica. However, the temperature of densification has been 
reduced to 1150°C. 

Conclusions 
The second goal of this study, which was to achieve dehydration of monolithic 
xerogels, has been accomplished. All the absorption peaks and bands in the range from 
200 nm to 4400 nm due to the presence of pore water and surface hydroxyl groups were 
identified. Monolithic gel-glasses were routinely produced by this sol-gel method in 
conjunction with the carbon tetrachloride treatment. These completely dehydrated 
samples were able to reach and maintain a completely hydrophobic surface. Further 
evidence thai these samples were completely densified is supported by mercury- 
displacement density measurements, with a resulting average value of 2.2 g/cm 3 -the 
density of fused silica glass of Types I, II, III, IV. The optical properties of these fully 
dehydrated gel-glasses will be evaluated in Chapter 5. 



158 



0.40 



0.20 ~ 



<d 0.00 

| 200 

CO 



o 

w 




800 



■g 0.40 



0.20 - 



0.00 



0.40 



0.20 



(b) 2 days 



200 



800 



(c) 4 days 



800 



0.00 I 1 I 

1 200 

CO 



.§0.40 

CO 



0.20 - 



0.00 



(d) 7 days 



1400 2000 

Wavelength (nm) 



2600 3200 



2668.8 nm 



♦- 



i ' i I ■ 



1400 2000 

Wavelength (nm) 



2600 3200 



2668.8 nm 




J I L 



1400 2000 

Wavelength (nm) 



2600 3200 



2668.8 nm 




I i I i I 



J L 



200 



800 



1400 2000 

Wavelength (nm) 
thickness 3.8 mm 



2600 3200 



Figure 4-19 Absorption curves of 1 050°C samples aged in air for various times. 



159 



0.40 



0.20 ~ 



0.00 




200 800 1400 2000 2600 3200 

Wavelength (nm) 



0.40 



0.20 ~ 



0.00 




200 800 



0.40 



0.20 - 



1400 2000 

Wavelength (nm) 



2600 3200 



0.00 




200 800 1400 2000 2600 3200 

Wavelength (nm) 



Figure 4-20 Absorption curve of 1 1 50°C sample aged in air for various times. 



CHAPTER 5 
OPTICAL PROPERTIES OF FULLY DEHYDRATED SIUCA GEL GLASS 



trodup lii; 

The initial approach towards producing high optical quality, pure silica gel glass 
monoliths via a chemically treated, thermal densification process was achieved, as 
described in Chapter 4. The purpose of this chapter is to investigate the optical 
properties of these samples and to compare them with those of traditional, high-purity, 
commercial type III and type IV silica glasses. 

Silica glass, whether produced by the traditional method or by the low- 
temperature sol-gel route, can be described as a solidified supercooled silica liquid of 
randomly packed silica tetrahedra in which a relatively stress free, short-range- 
ordered structure has been formed, as discussed in Chapter 3 Section IV. This solid 
appears to have a complete lack of periodicity and a tendency to "order" only in the sense 
that a few silica tetrahedra are fairly tightly packed together with a statistical 
preference for a particular interatomic distance, as indicated by x-ray scattering. For 
optical applications, silica glass usually is required to be an amorphous, isotropic, 
homogeneous, transparent, dielectric, insulating material. 

An ideal silica glass, defined as silica glass without nonbridging oxygen bonds or 
cation or anion impurities, does not exist in the real world. However, like an ideal gas an 
ideal silica glass can be approached. The sol-gel fabrication technique developed herein 
is a step forward to this goal since the new low-temperature route results in no 
absorption loss due to hydroxyl (OH") ions and minimal other ionic impurities in the 
ppb range. 



160 



161 

Interactions between electromagnetic radiation and glass, based on both quantum 
mechanics and classical treatments, has been well established in the literature [80, 
100, 101] and is the basis for interpreting the results presented in this chapter. 

Optical properties of gel glass are determined not only by intrinsic chemical 
aspects (e.g. electronic energy gap, interatomic bond strength, ionic mass, and impurity 
levels), but also by extrinsic physical aspects of the processing (e.g. thermal history, 
thermal gradients, structural arrangement, and degree of isotropy) developed during 
densification process. 

The physical properties of a glass are always interrelated; for example, molecular 
vibrations are responsible for light absorption, resonance, heat dissipation, 
fluorescence, phosphorescence and thermal expansion [102]. Refractive index is a 
function of density and electronic polarizability, etc. The optical properties to be 
examined in this chapter include vacuum ultraviolet (VUV) transmission, ultraviolet 
(UV) transmission, visible (VIS) and near infrared (NIR) transmission, infrared (IR) 
spectra, index of refraction (n), and dispersion (i>). In addition, the optical quality of 
the gel silica monoliths is tested by measurements of homogeneity, stress birefringence, 
striae, bubbles, inclusions, and impurities. This information along with coefficient of 
thermal expansion (CTE), density and microhardness (Knoop hardness, DPN) data are 
used to compare and characterize the gel-silica glasses. 

Literature Review Regarding Optical Properties of Silica Glass 
Classification of the ultraviolet cutoff wavelength of commercially available high 
purity fused silica has been made by Sigel [72]. He suggests that the location of the VUV 
(vacuum ultraviolet) absorption edge can be attributed to three factors: (a) a 
completely stoichiometric Si-0 network, with its strong O-Si-0 bridging bonds, which 
provides the minimum absorption wavelength at about 150 nm; (b) a small amount of 
terminal Si-0 bonds (e.g. silanol groups), also called non-bridging oxygen (NBO) bonds, 



162 

which determines the degree of shift to higher wavelengths in the 150 nm to 200 nm 
range; (c) significantly higher wavelength shifts, from 200 nm to 350 nm, which are 
induced by impurities (e.g. transition elements, alkali, alkaline-earth and halogen 
elements) in the ppm range, as listed in Tables 1-1, 1-2 and shown in Figure 1-1. 

Refinement of the sol-gel precursor, for example TEOS (tetraethylorthosilicate) 
reduces the metallic impurities to a minimal ppb level, as listed in Table 5-1, which 
makes it possible to produce a glass having a very high quality of light transmission in 
the VUV and UV. The elimination of physical and chemical water (also considered 
impurities) associated with the gel has been described in Chapter 4. An absolutely 
impurity-free silica glass should exhibit a VUV absorption edge of approximately 150 
nm, as indicated in factor (a) above. 

Silica glass is capable of being used as an "optical window" between the vacuum 
ultraviolet and infrared absorption (160 nm to 4400 nm) regions. The subregion from 
600 nm to 1100 nm is the portion of the electromagnetic spectrum of interest for 
present day long distance optical fiber communication systems [103]. However, the 
"window" from 600 - 1100 nm is not usually perfect since a number of material 
absorption and scattering losses are present. Loss mechanisms, including fundamental 
UV and IR absorption tails, overtone and combination peaks of hydroxyl groups, and 
Rayleigh scattering are shown in Figure 5-1. 

Rayleigh scattering is due to density and compositional variations in the material. 
Today, type IV silica is developed and produced commercially for making optical 
wavequides (fibers) which require extremely low signal loss for long distance use in 
optical communication cable systems. The best quality of silica optical fiber (type IV) 
has been achieved with an internal attenuation value around 0.2 dB/Km (10 dB = 1 OD 
optical density absorbance = 10% transmission, OD = Log [lo/l] where lo is the incident 
intensity, I is the transmitted intensity) at 1550 nm in single-mode operation [see p. 
32 in ref. 83]. Silica gel-glass optical fiber with no hydroxyl groups and minimal 



163 





Table 5-1 








Impurity levels in TEOS (ppb). 






Al = 20 


Li < 10 Cu <10 


Fe 


= 1000 


Ca =50 


Mn < 1 Ti < 20 


Mg 


= 100 


Co < 50 


Na =30 Cr < 50 


Zn 


<50 


K = 100 


Ni < 100 













164 



E 

m 

■o 



§ 3 

c 

< 



4 ~ 



2 - 








700 800 900 1000 1100 1200 1300 1400 1500 1600 

Wavelength (nm) 



Figure 5-1 Typical spectral loss curves of silica optical fibers. 



165 

metallic ion impurities has been made by Susa showing a low-loss 5.9 dB/km at 850 nm 
[104]. The gel-glass monoliths developed herein should have equivalent or even better 
quality. 

The infrared absorption (IR) spectrum of a glass can be used as a tool to understand 
the chemical composition, molecular vibrations and the molecular bonding within the 
material. According to both classical and quantum theories, as two atoms with partially 
filled outer electron orbital approach each other the energy will either increase or 
decrease. For example, if two outer electrons, one from each atom, have parallel 
oriented spins, the result will be repulsion and the energy will increase. The closer they 
move together, the higher the repulsive force, consequently, the atoms will move apart 
resulting in an antibonding molecular orbital arrangement. If the spins are antiparallel, 
the result will be attraction and the energy will decrease. The lowest energy is achieved 
when the bonding takes place as the two orbitals overlap and electrons are shared by two 
atoms. If the atoms come too close together, the repulsive force rises due to two 
positively charged nuclei. The energy curves of this bonding (Morse curve) and 
antibonding system is shown in Figure 5-2 [see p. 381-408 in ref. 102]. 

Atoms in a molecule can vibrate harmonically but not symmetrically with the 
atoms moving together and moving apart as a stretching mode. This vibration of a bonding 
molecule can be expressed as a horizontal line at a certain temperature within the 
bonding curve of Figure 5-3. As shown in quantum theory, the vibration levels are not 
continuous and always have many rotational levels associated with a vibration level and a 
lowest possible energy level, called zero point energy, must exist even at 0°K. The 
higher energy ievei of a stretching vibration results in a larger average displacement 
between two nuclei. Any vibration energy higher than the destruction of the bonding 
energy will move atoms apart. The vibration frequencies of all molecules is so low that 
the energy involved is too small to interact directly with visible light; however, the 
absorption due to vibrational transitions between the vibrational ground state and 



166 



+ 



E> 
<B 

LU 



antibonding 
(parallel spin) 



interatomic distance 




bonding 
(antiparallel spin) 



zero point energy 



Figure 5-2 The energy curves of the antibonding and bonding molecular orbitals. 



167 



+ 

n 



©Q 

UJ 




zero point energy at 0°K 



Figure 5-3 The vibration levels at various temperatures. 



168 

excited states are found in the energy range of the infrared spectrum, at least for silica 
and silicate glasses. 

A comprehensive simplified diagram of the molecular vibrational energy levels in 
a ground state Morse curve and two excited state Morse curves is shown in Figure 5-4. 
This diagram indicates that the IR absorption could possibly result in a series of possible 
energy transformations in a silica glass which include resonance, heat dissipation by 
internal conversion, fluorescence, heat dissipation by interstate crossing, and 
phosphorescence. 

Because of the asymmetry in the potential energy curve (Morse curve), the mean 
position of the center of mass of the vibrating atoms will be displaced as the amplitude of 
vibration increases, as shown in Figures 5-3 and 5-5(a), resulting in thermal 
expansion of the material. The thermal expansivity is determined by the asymmetry in 
the potential energy curve, and the deeper the minimum the more symmetrical is the 
curve near the bottom, as shown in Figure 5-5(b). A strong bonding material has a 
deeper valley of higher symmetry which results in a smaller thermal expansion. 

D. G. Holloway, in his book The Physical Properties of Glass [see p. 36-41 in ref. 
63], states that "silica glass shows an anomalous thermal expansion behavior: the 
thermal expansion coefficient (CTE) for this glass is very much lower than for quartz 
(80 x 10" 7 parallel to axis 134 x 10 -7 perpendicular to axis), and it becomes negative 
below about -80°C. This unusual behavior may be related to the very open structure of 
the network, since the density of quartz is 2.66 g/cm 3 and silica glass is 2.20 g/cm 3 , 
and the consequent predominance of vibrational modes involving displacements of the 
silicon and oxygen ions transverse to the bond direction". 

It Is reasonable to assume that the large interatomic space in vitreous silica 
partially accommodates the dimensional increase with temperature due to stretching 
vibrations and this reduces the thermal expansion effect. From the point of view of a 
classical spring model, it is relative easier to bend than to stretch a spring. 



169 



% 
§ 

•5 

© 
> 

I 

cc 




Figure 5-4 Possible energy transformation in a glass. 



170 



(a) 



(b) 



§ 

Q> 
C 
HI 



g 

o 

c 

UJ 



large 

thermal 

expansion 

interatomic 
spacing 




weak bonding 



small thermal 
^expansion 




Strong bonding 



Figure 5-5 Thermal expansion depends on bonding strength 



171 

Consequently, the stretching vibration of a Si-O-Si bond is at a higher energy, at around 
1130 cnr 1 than the rocking vibration which is at 480 crrr 1 . 

As the temperature decreases below a certain point (T c ) the stretching vibration 
can no longer contribute to a dimensional change due to the onset of symmetry of the 
Morse curve, as shown below point (T c ) in Figure 5-6(b) curve (1). 

It is also reasonable to propose that as the temperature continuously decreases 
below T c the strong transverse bending vibration of silica starts to reduce its amplitude. 
This consequently increases the interatomic distance as shown in Figure 5-6(b), curve 
(2), and thus changes the dimensions as shown in Figure 5-6(a) and curve (3) of 
Figure 5-6(b). 

The corresponding coefficient of thermal expansion curve of vitreous silica is 
shown in Figure 5-6(c). Thus, both the large interatomic space and the decrease of 
amplitude of bending vibrations with temperature below T c contribute to the unusual 
thermal expansion behavior in silica glass. As will be shown later, this "anomolous" CTE 
behavior of vitreous silica is dramatically different for a sol-gel derived silica. 

The optical properties are not solely determined by the chemical composition of the 
silica gel glass, but are also influenced by the densification procedure. Since the 
refractive index of glass is related to both chemical composition and density, it can be 
altered by changes in two interrelated intrinsic properties, the electronic 
polarizabilities of negatively charged chemical species (e.g., oxygen ion, chlorine ion) 
and density. The electronic polarizability, <x e , is an inverse function of the 
electronegativity [105]. The electronegativity of an oxygen ion (0~ 2 ) is higher than that 
of a chlorine ion (CI" 1 ). Consequently, the polarizability of a chlorine ion is higher than 
that of oxygen ion. 

The index of refraction (n) is proportional to the summation of the 
polarizability of all the chemical species in a glass. Because of the higher polarizability 
of the anions, n is primarily dependent on the summation of the anionic 



172 



(a) 



stretching 
vibration 



atO°K 



(b) 




& 

LU 



— vibration center 



this portion might 
contribute to 
negative CTE 



at T>0°K 




bending 
vibration 
model C 

atO°K 



spring 
models 



(c) uj 
o 

0°K 



large silica interatomic 
space acommodates 
the bending vibration 



curve 2: bending vibration 
t~~ contributes to negative CTE 



- curve 3 ■ curve 1 + curve 2 

curve J |_ stretching vibration 
! contributes only to positive CTE 




Temperature 



zero point energy at 0°K 



Figure 5-6 A proposed classical spring model and thermal expansion curves of silica glass 



173 

polarizabilities. This relationship is described by the Lorentz - Lorenz equation: 
a= 3 e (n 2 - t)M/[N (n 2 + 2)p] (1 ) 

Therefore n is directly proportional to a as the other items in the equation above 
are constants; i.e. 

N is Avogadro's number 

e is the dielectric constant of vacuum 

M is the molecular weight of silica 

p is the density of silica 
The second important effect to consider is density. Since a is constant for silica, on 
condition that other anionic impurities are negligible, the density is proportional to (n 2 
- 1)/(n 2 + 1). Thus, as density increases the index of refraction increases. 

Susa, Matsuyama, Satoh, and Suganuma at Hitachi Ltd. Japan [106], reported on 
the effect of chlorine content in sol-gel derived silica glass. They observed that the 
refractive index of the gel glass increases in proportion to the chlorine content. These 
findings are correct according to the larger chlorine ion polarizability discussed above. 
However, in their report the chlorine content of selected gel glass samples was 
determined by nephelometry, which is a method to measure the concentration of a 
suspension or substance or a second phase (e.g., bubbles, inclusions) by comparing the 
brightness of light passed through a sample with that passed through a standard and is 
incapable of directly measuring a colorless ionic solution (e.g., NaCI in water, or CI' 1 in 
glass). All the samples they produced had been heated to 1300°C, consequently, those 
with a higher initial chlorine (CI" 1 ) ion concentration were likely to have 
proportionally freed more chlorine gas (Cfe) to create more closed micropores. This 
structural change would decrease the brightness of incident light in the measurement and 
lower the apparent density. In addition the equilibrium of 2CI" 1 <— -> CI2 + 2e" 1 
should be a constant, K = [Cl2]/[CI" 1 ] 2 , at that temperature for all samples having 
various chlorine contents. Therefore, samples with a higher initial surface area were 



174 

expected to have a higher structural chlorine (CI' 1 ) residual attaching to the silica 
matrix which would contribute to a higher refractive index (freed CI2 gas which boiling 
point is -34.6°C, the index of refraction is 1.000768). It is thus reasonable to conclude 
that the measured refractive index in the report of Susa, et al. was mainly proportional 
to the concentration of micropores rather than to that of chlorine (CS~ 1 ) and the obtained 
apparent density decreased as the chlorine gas, CI2, content increased. 

According to the Lorentz - Lorenz relation, the refractive index is linearly 
proportional to the true density of silica (Si02) as shown in Figure 5-7 [107]. 
Consequently any changes in short-range-ordering, crystallization or structural 
transformations of vitreous silica that increases the density will also increase the 
refractive index. The true density can be varied in a sintering process by controlling the 
thermal history in the glass transition range of temperatures. Unfortunately, such phase 
changes or structural rearrangements in small scale (below 2 wt.%) is unable to be 
detected using x-ray diffraction. In addition, a small amount of a second phase 
(microvoids) found in gel dried at 160°C using a microscope, is x-ray undetectable by 
x-ray diffraction. 

The true density of a dehydrated gel glass with microvoids and closed micropores is 
difficult to determine. The apparent density, which has a value around 2.184 gm/cm 3 
comparing to 2.202 gm/cm 3 ((2.202 - 2.184)72.202 - 0.8 wt.%) of fused silica, 
shows the effect of a very small volume fraction of micropores. Thus, differences in 
refractive index can be due to either an increase of chlorine content or an increase in 
density. When both factors are present it requires a measurement other than 
nephelometry to separate them. 

The index of refraction of a material usually decreases as the wavelength (X) of 
light increases. This change with wavelength is called the dispersion of the index of 
refraction and is defined as dn/dX. However, most practical measurements are made by 
using the index of refraction at fixed wavelengths at the yellow helium d line (587.0740 



175 



c 
o 

'o 

CO 

CD 



x 

Q> 
"D 

c: 



1.56 " 












/ 






I 








/ 4 


1.54 




























* 




1.5<: 






























1.50 








^r 
























1.40 






/ 3 

/ I 












- 1 : Corning 7940 glass 

2 : Tridymite 

3 : Cristobalite 
4 : Quartz 










/°2 




1.46 






1 














1.44 " 




















1.42 " 


r^ 




! 




1 







2.0 



2.1 



2.2 



2.3 



2.4 



2.5 



2.6 



2.7 



True Density (g/cc) 



Figure 5-7 Index of Refraction versus True Density 



176 

nm), the blue hydrogen f line (486.1337 nm), and the red hydrogen c line (656.2725 
nm). The numerical difference between the two indices of refraction at the f and c lines 
is called the mean dispersion (nf - n c ). The ratio (nf - n c )/(nd - 1) is the dispersive 
power. Its inverse is called Abbe's value, (n d - 1}/ (nf- nc). The dispersion of the gel 
derived silica glasses developed herein is described in a later section. 

Examination of the homogeneity of a piece of glass for optical applications is very 
important since the wavefront of incident light can be distorted by any variations of 
index of refraction in a nonhomogeneous glass. Index variations can be caused by 
localized thermal gradients which induce density gradients. Such inhomogeneities can 
result from either an improper sintering process or impurity fluctuations within the 
densified glass. Consequently, careful control of both densification and the dehydration 
process is necessary to provide a uniform refractive index throughout the gel glass body. 

Interferometry is a precision measurement which can be used to examine the 
quality of a material surface, to metrology, to the alignment of optical and mechanical 
components, or to examine the differences in optical path length (S - Ln where S is 
optical length, L is sample thickness, n is index of refraction) in a glass. The path 
difference, T, is the difference between two such path lengths, T = S2 - Si = L (n2 - 
n-|), where thickness L is constant. From Young's double-slit experiment, T can be 
expressed as T =mX where m is the order of interference or the number of fringes. 

By rearranging the two equations above we have m = L(n2 - n-|)/X[see p. 187- 
204 in ref. 100]. Consequently, for a piece of glass with two perfectly parallel surfaces 
and constant thickness, any internal irregular variation of refractive index results in an 
irregular fringe shift pattern on an interferogram. An interferogram always shows 
alternative dark bands and white bands with one fringe corresponding to one pair of 
white and dark bands. This is the method used to examine optical homogeneity in the 
densified gel-glass samples. 



177 

Internal stress in glass can be produced by many factors such as mechanical stress, 
thermal quenching, phase separation, crystallization, etc.. For example, If a piece of 
glass quenched from high temperature has residual stresses (tension and compression) 
and if light is propagated through such a glass the difference in refractive index between 
the regions of stress results in stress birefringence. The birefringence is defined as the 
numerical difference between the two refractive indices (e - co) or a measure of path 
difference, T, per sample thickness, L. Consequently, the birefringence can be expressed 
as the retardation of light which is e - co = T/L where T is 85t/2n, 8 is the phase 
difference, and k is wavelength of incident light [see p. 339-341 in ref. 100]. 

To examine the stress induced birefringence the addition of two polarizing filters 
and a rotatable stage converts a laboratory microscope into an polarizing microscope. If 
a piece of strain-free glass is placed on the stage between these two crossed polarizers, 
the glass remains dark no matter how the stage is turned; such a glass is isotropic. 
Anisotropic glass, in contrast, has four positions of maximum extinction, 90° apart, 
when the stage is rotated. The phase difference, 8, can be determined by measuring the 
angle of rotation to give compensation. 

To be optically useful a silica glass must be able to transmit electromagnetic waves 
efficiently within the region in which it is to be used, i.e., it must exhibit a very low 
scattering. For uniform interaction between light and glass, the glass should not only 
have a homogeneous index of refraction but also be free of excessive striae, bubbles and 
inclusions. In traditional melt glasses, these three types of defects are often found if the 
thermal processes are inadequate. Striae result from incomplete homogenization and 
high temperature mixing in the liquid phase prior to casting. Bubbles from chemical 
reaction of raw materials and inclusions from unmelted high-temperature impurities on 
the scale of micrometers to millimeters, and even larger, are formed and trapped during 
an improper melting and cooling process. Consequently, it is necessary to examine 
whether such imperfections are present in a gel glass. 



178 

Finally, impurities such as alkali and alkali earth elements, transition-metal 
elements, and halogen elements terminate the bridging oxygen bonds, create light 
interaction centers and degrade the optical performance of a silica glass. All of the above 
physical and structure factors must be determined to characterize the quality of the gel- 
silica glass produced herein. 

Experimental Proned|»re 

Glass Fab, Inc. of Rochester, New York was selected to evaluate these first 
generation ultrapure silica gel glass, produced as described in Chapter 4, as potential 
optical components. They were contracted to perform optical performance 
characteristics and properties tests on six gel-silica glass samples. Several 
commercially available, high quality type III optical silica glasses were used for 
comparison. The samples were three high purity fused silica samples (Corning 7940) 
and three synthetic optical quartz samples (NSG quartz - type ES). Comparative optical 
transmission and stress birefringence data were also obtained at the Advanced Materials 
Research Center (AMRC) of University of Florida. The tests performed on these samples 
are listed in Table 5-2. 

Prior to Glass Fab's transmission testing they polished all samples simultaneously 
to 0.5 wavelength of red helium light (706.5188 nm). After polishing, the samples 
were tested for flatness on two surfaces to 0.5 wavelength flatness. Samples were then 
cut into 20 mm squares and two surfaces were polished to a 90 degree angle. 

Vacuum ultraviolet (VUV) transmission tests, in the 160 nm to 200 nm range, 
were performed by Glass Fab on an Acton Research Corporation, 0.2 meter (focal 
length), Model VM-502 with an uncertainty of ± 2%. 

Transmission in the UV-VIS-NIR range, 200 nm to 2600 nm, was measured by 
Glass Fab using a double-beam Perkin-Elmer spectrophotometer, slit width 2-10 nm, 
with an uncertainty of ±1%. Transmission measurements were made in the 186 nm to 



179 



Table 5-2 
Physical property measurements on fully densified ge!-silica glasses 

and fused silica glasses 

Test Number of Samples measured Source 



Optical tests: 
Transmittance 

(1) Vacuum UV 

(2) UV-VIS-NIR 

(3) IR 

Refractive index 

Dispersion 

Homogeneity 

Striae 

Stress birefringence 

Bubbles and Inclusions 

Impurity 

Thermal and mechanical test: 

Coefficient of thermal 
expansion 

Specific gravity 

Knoop hardness 



6 gel glass samples 
6 control samples 



6 gel glass samples 
6 control samples 
6 gel glass samples 
6 control samples 
6 gel glass samples 
6 control samples 
6 gel glass samples 
6 control samples 
6 gel glass samples 
6 control samples 
6 gel glass samples 
6 control samples 
1 gel glass sample 



Glass Fab 



(a) 1 gel glass sample 

(b) 2 gel glass samples 

3 gel glass samples 
2 control samples 
1 gel glass sample 
1 control sample 



Glass Fab 

Glass Fab 

Glass Fab 

Glass Fab 

Glass Fab 

Glass Fab 
North Carolina 
State University 



Penn State University 
University of Arizona 

Corning Engineering 
Lab Services 
Corning Engineering 
Lab Services 



180 

3200 nm range at the AMRC on approximately 50 unpolished gel-silica glass samples 
using a double-beam Perkin-Elmer Lamda 9 UV/VIS/NIR spectrophotometer, Model 33, 
slit width 1 nm, with an uncertainty of ± 1%. 

Infrared transmittance was also measured in the 2500 nm to 5000 nm range by 
Glass Fab using the spectrophotometer previously mentioned. 

Refractive indices were measured by Glass Fab on a Pulfrix Abbe Refractometer 
using four special light sources, isolating the six spectral lines at which the tests were 
conducted, as listed in Table 3. Calibration was accomplished by use of a standard index 
sample certified by the National Bureau of Standards (NBS) accurate to ±1 x 10" 5 . 
Dispersion (dn/dk) values were calculated from refractive indices at different testing 
wavelengths in according with the Abbe's value vd = (nd - 1)/ (nf - n c ) defined in 
Section II of this chapter. 

Homogeneity of the gel glass and control samples was checked for wavefront 
distortion by Glass Fab on a Zygo Zapp Interferometer. Samples were examined and then 
additionally tested using oiled-on master plates to eliminate any effects of polishing. 

Striae tests were made by Glass Fab using a pin hole arc lamp to project an image 
10 time size onto a projection screen. In this test any striae in a glass appear as fine 
lines on the screen. 

Stress birefringence tests were performed by Glass Fab using a Fridel Polariscope, 
Polarmetrics Model 35 polarimeter. Prior to cutting, polished samples were examined 
in two directions. Any visible strain appeared as a field change (a twist of the polarized 
length). Using a rotating eyepiece, the field was rotated until the field change was 
reversed. This angle change was used to determine the retardation level, R, (strain) 
using the formula: R « 3.3 A/T, where A is angle of rotation to give compensation, and T 
is thickness of sample. 



181 



Table 5-3 
Optical dispersion wavelengths 



Designation 


Wavelength 


(nm) 


Spectral Line 


r 


706.5188 




red helium line 


c 


656.2725 




red hydrogen line 


d 


587.5618 




yellow helium line 


e 


546.0740 




green mercury line 


f 


486.1327 




blue hydrogen line 


h 


404.6561 




violet mercury line 



182 

Stress birefringence was qualitatively determined at the AMRC on "as cast", 
partially dense (-60%) and fully dense gel-silica glass samples using two plane 
polarized laminated plastic sheets. 

Bubbles and inclusions were examined by Glass Fab using a Nomarski phase- 
contrast microscope. In this test the magnification is set at 400X and the resolution is 1 
micrometer. The number of bubbles and inclusions were counted in a volume 0.02 mm 3 . 
Volumes were randomly selected in each quadrant and at the center of the samples. The 
sampling volume was defined by a 0.5 mm diameter field of view and a 0.1 mm sweep 
inside the sample about 0.5 mm from a polished face, as shown in Figure 5-8. 

Impurity levels were measured in a gel glass sample by the Department of Nuclear 
Engineering, North Carolina State University using a neutron activation analysis 
technique [108] to measure the number and energy of gamma and x-rays emitted by the 
radioactive isotopes produced in the sample matrix. This method involves irradiation of 
the test sample with thermal neutrons from a nuclear reactor at a selected time period. 
Quantitative analysis was obtained by comparing the number of characteristic x- or 
gamma rays detected from the gel glass with the number determined for appropriate 
standards. 

Other physical properties measured included coefficient of thermal expansion, 
specific gravity and knoop hardness. 

Coefficient of thermal expansion (CTE) values from room temperature to 773 K 
were measured on one gel glass sample by the Materials Research Laboratory, 
Pennsylvania State University using a laser speckle dilatometer and on one gel glass 
sample by Orton Jr. Ceramic Foundation, Ohio, using an automatic recording dilatometer. 
The new laser speckle method [109] was based on the movement of a laser beam 
reflected from a probe bar caused by thermal expansions of the sample and reference 
rods as shown in Figure 5-9. The reflected laser beam was observed with a small 
photodetector. A minute change in the laser beam position thus results in a precision 



183 



gel glass sample 



CJ"_"_W polished face 



»j 0.5 mm 



0.02 mm 3 
each quadrant 




0.5 mm 



Figure 5-8 Volume for counting "bubbles or stars" 



184 



Furnace 



Probe 



Laser Beam — — - 



n: 



Sample 



XT 




Reference 



Platform 



Aperture and 
Photodectector 



Figure 5-9 Schematic diagram of a laser speckie dilatometer 



185 

thermal expansion measurement. The Orton dilatometer was laboratory calibrated for 
accuracy against platinum to help insure precise measurement of thermal length changes 
of the gel glass sample. 

CTE values from 4 K to 473 K were obtained on two gel glass samples with five 
control samples by the Optica! Science Center, University of Arizona using a low 
temperature laser interferometer dilatometer. 

Precision apparent density measurements were obtained on three gel glass samples 
with two control samples (Corning 7940 and NSG ES fused silicas) by Corning 
Engineering Lab Services using a simple deionized water displacement buoyancy method. 

Knoop Microhardness values of one gel glass sample and one control sample were 
determined by Corning Engineering Lab Services using a 100 gram loaded knoop 
hardness tester in accordance with ASTM C-730 testing procedure [110]. 

Results and Discussions 

Transmission data of the six silica gel glass samples are separated into three 
sections, according to their wavelength testing ranges, and compared with the traditional 
glass control samples. The results shown in Figures 5 - 10, 11, 12 and Table 5-4. The 
gel glass samples demonstrate a uniformly high transmittance in the 200 nm to 2600 
nm range. In the vacuum ultraviolet region at 165 nm the gel glass has five times the 
transmittance of the Corning 7940 sample and 2.5 times that of the NSG-ES samples. 
The flat transmission spectra of the gel glass is evidence that cation impurity 
contaminations have been minimized and water has been eliminated. In contrast, 
transmission spectra for both control samples show significant water absorption peaks 
at the wavelengths of 1370 nm and 2200 nm. 

The gel-silica glass also shows substantially greater transmission than the 
traditional type III silica glasses when tested in the far IR range from 2500 nm to 5000 
nm, as shown in Figure 5-12. Corning 7940 and NSG-ES fused silica samples show 0% 



186 



100 



80 - 



CO 

E 

CO 

e 

03 



60 - 



40 - 



20 - 



160 




n gel silica 
O Corning 7940 
NSG-ES 



170 180 190 

Wavelength (nm) 



200 



Figure 5-10 Vacuum ultraviolet transmission of optical silicas 
sample thickness: 3 mm. 



187 



100 



80 - 



§ 60 



E 

CO 

S 40 ■ 




20 - 



gel silica 
Corning 7940 
NSG-ES 



200 600 1000 1400 1800 

Wavelength (nm) 



2200 2600 



Figure 5-1 1 UV-VIS-NIR transmission of optical silicas 
sample thickness: 3 mm. 



188 



100 



-ft 



c 
o 

to 

"E 

0) 

£= 

03 



60 - 



40 - 



t 
\\ 

w 

\ 1 

\i 

I I 
: 

If 

i 

s 



£0 



1 1 



1 1 
'I 

;/ 



; 

if 

i 

n 
I 



2500 



3000 




gel silica 
Corning 7940 
NSG-ES 



3500 



4000 



4500 



5000 



Wavelength (nm) 



Figure 5-12 Infrared transmission of optical silicas 
sample thickness: 3 mm. 



189 



Table 5-4 
Vacuum ultraviolet transmission data 

Sample Transmission (%) at Wavelength of 

ID No. 165 nm 170 nm 176 nm 180 nm 190 nm 200 nm 

Gel glass test samples 3 mm thick: 



Q27 


13.9 


65.0 


80.0 


83.0 


84.5 


85.5 


N 34 


14.0 


63.0 


78.0 


80.5 


82.5 


83.5 


P 37 


15 .0 


64.0 


78.5 


80.5 


83.0 


84.0 



Corning # 7940 Control Sample, 2 mm thick: 

CGW-2(a) 08.0 60.0 76.0 83.0 88.0 91.0 

Corning # 7940 Control Sample, Converted to 3 mm thickness: 

CGW-2 (a) 03.0 47.0 67.0 77.0 82.5 87.0 

From reference, reflection loss per single surface: 

R Loss (b) 05.7 05.5 05.3 05.1 04.9 04.7 

Notes: 

a) As noted, the Corning #7940 sample measured was 2 mm thick compared to 
the 3 mm thick gel-glass test samples. This data was converted to 3 mm thickness 
for comparison. 

b) Reflection losses shown are based on published data for fused silica available 
from Glass Fab, Inc.. and is presented for reference only. 



190 

transmission at the hydroxyl group absorption peak of 2730 nm, compared to 93% 
transmission of gel-silica glass. 

Precise refractive index measurements of the silica gel glass and control silica 
samples are fisted in Table 5-5 as a function of measuring wavelength. The mean value 
and its standard deviation of the index at each wavelength is also shown in Table 5-5. A 
summary of the dispersion data is given in Figure 5-13. The size of data points in Figure 
5-13 represents the variation of the data in Table 5-5. 

The reference index n<j and the calculated Abbe constant for each type of silica is 
listed in Table 5-6. The variation in refractive index (d-line), from sample to sample 
of gel glass, indicates that this characteristic is related to variations in thermal 
processing. 

A homogeneity test on one gel glass sample, Figure 5-14, shows an approximately 
0.863 wave peak to valley (P/V) ratio wavefront distortion in the inner 25 mm area, 
which compares to the Corning samples with a 0.529 P/V wavefront distortion (equal to 
the power of the polished surface). However, further examination of the gel glass shows 
a roll off with 4 to 5 waves P/V significant distortion at the outer 5 mm. No strain was 
evidenced on the ge! glass edge; therefore, it is clear that the inhomogeneity in the 
material and it's edges is strictly due to changes in refractive index within the material. 
This variation is quite pronounced and consistent at the ends, but appears to be better in 
the middle. The localized variation in refractive index is a result of density gradients 
caused by thermal gradients and/or chlorine impurity gradients. 

No striae were visible during the striae test, indicating that there are no surface 
irregularities, capillaries, or localized structural defects within the glass. This was 
true for the six gei glass samples as well as the six control samples. 



191 



Table 5-5 
Refractive index measurements of the gel-silica glass and fused silica glasses 



Test No. 
ID No. 



Index of refraction (n): 
d e f 



Corning # 7940 Control Samples: 



1 ) CGW-1 

2) CGW-2 

3) CGW-3 



1.45518 
1.45516 
1.45517 



1.45639 
1.45638 

1.45639 



1.45848 
1.45848 
1.45848 



1.46010 
1.46010 
1.46010 



1.46316 
1.46317 
1.46316 



1.46965 
1.46968 
1.46965 



Statistical Value: 



(±) 



1.45517 
0.00001 



1.45638 
0.00001 



1.45848 
0.00000 



1.46010 
0.00000 



1.46316 
0.00001 



1.46966 
0.00002 



NSG "ES" Control Samples: 



4} 
5) 
6) 



NSG-1 

NSG-2 
NSG-3 



1.45516 
1.45517 
1.45514 



1.45638 
1.45638 
1.45636 



1.45847 
1.45848 
1.45846 



.46009 
.46009 
.46008 



1.46315 
1.46315 
1.46315 



1.46965 
1.46965 
1.46967 



Statistical Value: 



(±) 



1.45516 
0.00001 



1.45637 

0.00001 



1.45847 
0.00001 



1.46009 
0.00001 



1.46315 
0.00000 



1.46966 
0.00001 



Gel-Glass Samples: 



7) 
8) 

9) 

11) 
12) 



N34 
Q34 
Q27 
Q11 
Q30 



1.45978 
1.45936 
1.45983 
1.45994 

1.45979 



1.46102 
1.46061 
1.46109 
1.46119 

1.45104 



1.46317 
1.46276 
1.46326 
1.46334 

1.46319 



1.46483 
1.46443 
1.46448 
1.46501 
1.46485 



1.46797 
1.46057 
1.46764 
1.46817 
1.46800 



1.47464 
1.47426 
1.47435 
1.47487 
1.47468 



Statistical Value: 



(±) 



1.45968 
0.00024 



1.46093 
0.00024 



1.46309 
0.00024 



1.46476 
0.00024 



1.46791 
0.00024 



1.47461 
0.00024 



192 





1.48000 








n gel silica 
• Coming 7940 
O NSG-ES 


^ 


1 .47000 


a N, 








c: 




\ X 








o 




\ x 








§ 




\ 












\ 








M— 




\s 








o 




N 








X 
C 


1.46000 
1 .45000 


I I 


I 


I 


1 1 1 



400 



500 600 700 

Wavelength (nm) 



800 



Figure 5-13 Dispersion data comparison of optical silicas. 



193 



Table 5-6 
Reference indices and Abbe values of silica glasses 



Test No. ID. No. 


Reference index (n^) 


Abbe Value (v^) 


Corning #7940, Control Samples: 




1 ) CGW-1 

2) CGW-2 

3) CGW-3 


1.45848 
1.45848 
1.45848 


67.72 
67.52 
67.72 


Statistical Value: 
(±) 


1.45848 
0.00000 


67.65 
0.11 


NSG-ES, Control Samples: 






4) NSG-I 

5) NSG-2 

6) NSG-3 


1.45847 
1.45848 
4.45846 


67.72 
67.72 
67.52 


Statistical Value: 
(±) 


1.45848 
0.00001 


67.65 
0.11 


Gel Glass Test Samples: 






7) N34 

8) Q34 

9) Q27 

10) P37 

11) Q11 

12) Q30 


1.46317 
1.46276 
1.46326 
1.46281 
1.46334 
1.46319 


66.64 
66.49 
65.90 
66.21 
66.38 
66.55 


Statistical Value: 
(±) 


1.46309 
0.00024 


66.36 
0.27 



194 




Figure 5-14 Homogeneity tests of silica gel glass sample and Corning #7940 control 

sample by Zygo Zapp Interferometer. 



195 

The stress birefringence test showed that through the faces of the six, 30 mm x 3 
mm (diameter x thickness), fully densified gel-silica glass samples no stress or strain 
could be measured. Through the ends (30 mm length) strain was observed which 
computed to 4 millimicrons (nanometers) per centimeter. For comparison, normal 
optical glass, per MIL-G-174 [111], should have less than 10 millimicrons per 
centimeter. The birefringence constant (R) of 4 nm/cm determined for the gel-silica 
glass is nearly equivalent to the values of 3.54 nm/cm and 5 nm/cm of Corning 7940 
and NSG-ES samples, respectively. 

The strain associated with partially densified gel-silica glass samples, using two 
plane polarized films, is shown in Figure 5-15. The strain present in partially 
densified gel-silica is eliminated by the densification process (Figure 5-16). This 
effect can perhaps be characterized as "precision annealling". 

The six control samples tested showed no evidence of bubbles or inclusions. All of 
the silica gel glasses exhibited bubbles approximately one micrometer in size. The so- 
called "bubbles" are really optical diffraction points. They appear more like "stars" than 
"bubbles". They are closed micropores created by the freed chlorine gas inside densified 
gel glass, as discussed in Section II of this chapter. Since the tested samples vary in 
quantity of these diffraction points, they seem directly related to thermal process 
parameters and should, therefore, be able to be eliminated by optimization of the 
thermai-chemical processing. Due to these defects the first generation gel silica samples 
would not acceptable for certain precision optical applications. Further analysis of the 
test results shows the spacing between points to be fairly homogeneous at a distance of 
about 75-125 microns. 

Impurity tests by neutron activation analysis show a significant chlorine content at 
about 0.1 wt.% with the other impurities in the ten to hundred ppb. No hydroxyl groups 
were detected. Except for the chlorine content, all impurity levels were below the levels 
of commercially available Types III and IV fused silica. 



196 




Figure 5-15 Observed strain in a partially dense gel-silica glass 



197 




Figure 5-16 Strain elimination in a fully dense gel-silica glass 



198 

The physical characteristics of silica glass, which generally include optical, 
thermal, and mechanical properties, are interrelated. For example, in the gel glass 
samples, having a large number of diffraction points or "microvoids", lower densities 
were measured, and lower Knoop hardness values were obtained. Also, the degree of 
inhomogeneily observed directly relates to the degree of refractive index variation 
measured. These daJa are supported by the Lorenfz - Lorenz relationship, which 
indicates that refractive index variation is the result of density gradients and/or 
chlorine gradients within the gel glass. 

Coefficient of thermal expansion (CTE) measurements from three testing 
laboratories are listed in Table 5-7. Figure 5-17 presents that data collected by the 
Optica! Sciences Center of the University of Arizona, courtesy of Dr. Steve Jacobs. 
Statistically, gel silica has a CTE value about two times lower and a more stable CTE over 
a wide range of temperatures than ail five of the commercial glasses tested. As discussed 
earlier the possible reason for the gel silica glass CTE behavior is the lower 
concentration of cation impurities and a larger intermolecular space between the silica 
structural units. 

Density measurements made on three gel silica samples and two control samples 
(Corning 7940, NSG-ES) are shown in Table 5-8. The densities of the gel glass 
measured, on an average, 0.016 g/cm 3 lower than the control samples. Micropores are 
responsible for these somewhat lower density measurements. 

Knoop hardness values were measured on one gel silica sample and one control 
sample (Corning 7940) with the results shown in Table 5-9. The gel silica measured 
lower than the control sample; however, the control sample measured significantly 
lower than it's published data. Corning Engineering Laboratory Services reperformed 
their tests and support those results. Though inconclusive, these data are an indication 
that the first generation gel silica has a lower hardness than fused silica which is 



199 



Table 5-7 
Coefficient of thermal expansion of fully dense gel silica (a)/°C 



Temp 

°c 


Temp 

°K 

150 
225 


Orton 


Penn 
State 


Univ. of 
Arizona 

2 x 10-7 
2 x 10-7 


25 


298 


0.4 x 10' 7 


-1.0 x 10-7 


2 x 10-7 


100 


373 


1.1 x 10" 7 


-1.0 x 10-7 


2 x 10-7 


200 


473 


1.4 x 10- 7 


3.1 X 10-7 


2 x 10-7 


300 


573 


1.9 x 10- 7 


3.1 X 10-7 




400 


673 


1.0 x 10- 7 


3.1 x 10-7 




500 


773 


2.4 X 10-7 


3.1 X 10-7 





200 



400 



300 



200 



100 



o 



UJ 



-100 




-200 







Zerodur 



Turning 7971 



J L__J_ 



i_J I L 



100 



200 300 400 

Temperature (°K) 



500 



600 



Figure 5-17 Coefficient of thermal expansion of gel silica compare with other fused glasses. 



201 



Table 5-8 
Density measurements of gel-silica and fused silica 



Test Sample 
Sample No. 



ID No. 



Density 

(Specific Gravity, gm/cm 3 ) 



8 


Q34 


2.1829 


9 


Q27 


2.1835 


10 


P 37 


2.1844 



Gel Glass Samples, Silica Material: 

No. 
No. 
No. 



Control sample, Corning #7940, Fused Silica: 

No. 1 CGW-1 

Control Sample, NSG ES Fused Silica: 

No. 4 NSG-ES 



2.2020 



2.2002 



For Reference Only: 

Following are the density (specific gravity) characteristics of various materials 

based upon published data, grams per cubic centimeter: 



1. Fused Silica, Corning #7940: 

2. Fused Silica (Synthetic) NSG-ES: 

3. Fused Quartz (Natural) NSG: 



2.202 (Same as measured) 
2.201 ( > measured value) 
2.203 



202 



TEST 
SAMPLE NO. 



Table 5-9 
Knoop hardness 



SAMPLE 
ID No. 



100 gm Load, Kg / mmH 2 



Gel Glass Sample: 
No.Q11 

Corning #7940: 
No. CGW-3 



456 



508 



Standard Deviation: 16.6 



Standard Deviation: 11.2 



For reference only: 

Following are the Knoop Hardness characteristics of various materials based 
upon published data (100 gm Load): 



1. Fused Silica (Synthetic) 600 - 630 

2. Fused Quartz (Natural) 590 - 620 

Notes: 

It should be noted that the Knoop hardness of the control sample of Corning 
fused silica measured lower than published data. Other Corning published data 
on the same material is lower than was measured. 



203 

consistent with the density data. It is believed that these low microhardness 
measurements are also related to the presence of micropores. 

A summary of the property data for the commercial dense gel-silica made by a 
modification of the process developed in this study is presented in Table 5-10. The gel- 
glass is designated as a type V silica for comparing with the other four types of 
traditional vitreous silicas. 

Conclusions 

Data from the characterization tests on the first generation of silica gel-glass 
monoliths were compared to commercially available control samples of high quality 
fused silica (type III). It was shown that the VUV-UV-VIS-NIR-IR transmission of gel 
silica glass is superior to that of fused silica, as observed from its broader transmission 
range approaching the theoretical value of ideal silica glass, 150 nm to 4400 nm. Gel- 
silica's broad transmission spectra having no absorption peaks is conclusive evidence 
that the major impurities, except chlorine, have been successfully reduced to ppb 
levels. The variation in index of refraction from sample to sample indicates that the 
chemical dehydration process and/or thermal process greatly influence the homogeneity 
of the gel-silica optica! material, and can thereby be controlled. Recent improvements in 
thermal processing have elemininated most of this source of variation, as indicated in 
Table 5-10. No stress birefringence or strain was observed, indicating that the samples 
were well annealled; also no striae were found. 

A significant number of microvoids were observed in the first generation gel silica 
glass samples. These inhomogenesties resulted in a somewhat lower apparent density and 
lower Knoop hardness value. Also density variations and/or chlorine gradients within 
the gel silica glass induced a refractive index gradient, which seriously distorted the 
incoming wavefront in the optical homogeneity tests. 



204 



Table 5-10 
Preparation and characteristics of five types of silica glass 

Type of Silica I II III IV V 



Electromeltec 


Flame-Fuse< 


i Hydrolyzed 


Oxidized 


Dense 


Quartz 


Quartz 


SiCI 4 


SiCI 4 


Gel-Silica 


Tradenames 










Vitreosil-IR a 


Homosil b 


7940 c 


Spectrosil WF a Gelsil f 


lnfrasil b 


NSG-OX 6 


1000 d 


7943° 






Optosil b 


Suprasil b 
NSG-ES 6 


Suprasil-W b 




Total Cation: 










(ppm) 30-200 


10-30 


1-2 


1-2 


1-2 


OH -1 group: 










(ppm) <5 


150-1500 


600-1000 


0.4-5 


<1 


CM: 










(ppm) 





100 


< 200 


< 1000 


UV 50% transmission: 










(nm) 212-223 


210-220 


165-188 


165-180 


165-168 


Thermal Expansion Coefficient: 








(x10 7 ) 5.4 


5.5 


5.5-5.7 


5.5 


2.0 


Bubbles and Inclusions: 










(#/in 3 ) 0-8 


0-5 


0-3 


0-2 





Strain: 










(nm/cm) 5-10 


5-10 


5-10 


10-40 


5 


Refractive index: 










(n d ) 1.458 


1.458 


1.458 


1.458 


1.458-1.463 


Dispersion: 










(\) d ) 67.8 


67.8 


67.8 


67.8 


66.4 -67.8 


Density: 










(g/cm 3 ) 2.21 


2.21 


2.20 


2.20 


2.20 



a = Thermal American Fused Quartz; Montville, NJ. 

b =Heraus Amersil; Sayrevilie, NJ. c = Corning Glass Work; Corning, NY. 

d = Dynasil; Berlin, NJ e ■ NSG quartz; Japan. 

f = Material Engineering & Science, University of Florida & GelTech, Inc.; Alachua, Fl. 



205 

The gel glass coefficient of thermal expansion is linear over a wide temperature 
range and lower than that of any previous fused silica. Less impurities and/or a larger 
intermolecular volume may account for both the low and anomalous thermal expansion 
behavior. 

Elimination of the density variations, microvoids and micropores can be 
accomplished with a finat optimization of the sol-gel process. More precise control of 
the thermal program and the dehydration technique in densification process will make it 
possible for gel silica glass to approach the theoretical optical performance of an ideal 
silica glass. 



CHAPTER 6 
SIUCA GEL OPTICAL FILTERS USING TRANSITION-METAL COMPOUNDS 



Introduction 

Large, monolithic pure silica gels have been made rapidly and reliably from 
tetramethylorthosilicate (TMOS) using drying control chemical additives (DCCA). In 
this chapter attention is shifted to using the TMOS-DCCA method to make silica gels with 
optical filter characteristics by introducing transition-metal ions into the transparent 
and colorless matrix. When the impurity is added, five incomplete but equal energy 
levels are split by the ligand field of the matrix. Ions with incomplete, split d electronic 
excitational and associated vibrational levels are responsible for absorbing light in 
characteristic ranges of wavelength in the gel matrix. Color observed in the gel glass 
containing transition-metal ions is the complementary color to the region of optical 
absorption due to these excitational and vibrational electron transitions. For instance 
Cr 3+ in a distorted octahedral ligand field of crystalline alumina (ruby) absorbs the 
violet and green-yeifow from the spectra, thus giving ruby its beautiful red color with a 
slight purple overtone. 

If an optical absorption is simply due to an electronic transition between two 
electronic levels, then the absorption bands should be very sharp. However, for glass 
containing transition-metal ions the hand widths are very broad. This implies that the 
associated vibrational levels of an excitational state interact with each other in the 
presence of lfo& ligand field, and also indicates that by altering the composition of the 
glass, the ligand field strengh can be changed. Consequently, variations of bonding 
strength may shift the absorption spectra and alter the band width. Unlike transition- 
metal ions, the ligand field is effectively shielded by the outer s and p orbitals of a rare 



206 



207 



earth element; thus the electron transitions taking place in inner f orbitals have less 
ligand field effects. The resulting absorption peaks are much sharper and comparable to 
that of a free ion. 

Color formation in glass arises from excitation of unpaired electrons in the d or f 
orbitals of the transition-metal ion or the rare earth element incorporated within the 
glass networks. Colors of transition-meia! ion doped materials are particularly subject 
to change by the variation of the coordination numbers and the splitting of the outer five 
d energy levels associated with the chemical bonding of the adjacent ions. Therefore the 
colors in such materials are described as resulting from specific chromophores, which 
are complex ions that produce a particular optical absorption effect [112]. In contrast, 
rare earth colorants depending on electron transitions in the inner f shell are much less 
subject to the local chemical environment of the coloring element; therefore, the change 
in color with local chemical bonding is minimal [113]. 

in this chapter are discussed the variations in colors and spectral absorption due 
to excitation of electrons in the silica gels and gel-glasses containing transition-metal 
ions. The optical properties are interpreted using the ligand field theory of 
chromophores. The theories of ligand field and molecular orbital transitions in d-shell 
colorants are reviewed and applied to the optical spectra of the chemically doped gel 
glasses. The transition metal elements investigated are Co, Cu, and Ni ion-doped silica 
gels and gel-glasses. The effects of thermal history are also studied. The results are 
compared with the spectra of the same elements in meit derived silicate and phosphate 
glasses. 

Review of L iterature 
Pure silica gel-glass has a very wide optical transmission range from the 
vacuum ultraviolet, 163 nm (7.60 eV, 61347 cnr 1 ) to the infrared, 4400 nm (0.28 
eV, 2272 cm -1 ) as described in previous chapters. The silica O-Si-0 bonding electrons 



208 



have a fully filled noble gas electron shell. Consequently, no incoming photon with lower 
energy than 7.6 eV can excite these strongly bonded electrons to higher quantum levels; 
as a result pure silica gel glass is transparent and colorless to human eyes. 

Coloring is one of the most important arts in human life. Artists since ancient 
China have tried successfully to preserve their masterpieces forever using vivid colors 
in porcelain ceramic glazes. The coloring constituents and molecular ratios of the glazes 
are trade secrets since they can only be developed by way of trial and error. 

The most common coloring ingredients found in ceramic arts are the transition 
metal ions characterized by an incomplete d electron shell, particularly V, Cr, Mn, Fe, 
Co, Ni, Cu. Rare earth elements, such as Nd, Er and characterized by an incomplete f 
shell, are less frequently used due to their cost and rareness. Insoluble metallic 
colorants such as Au are also used but will not be considered in this chapter since the 
chemistry is so dissimilar to that of the transition-metal ions. 

Ligand-field theory, which is a special case of the most general molecular orbital 
theory, is an alternative to crystal-field theory [114-116] to explain the color 
formation of transition metal doped silica glasses. In crystal field theory, bonding is 
treated as electrostatic, derived from the electric field of the ligands viewed as purely 
ionic species. Thus, in a crystal field method the chemical compound of a transition metal 
ion is considered as an aggregate of ions and/or dipolar molecules which symmetrically 
interact with each other electrostatically but do not exchange electrons. Consequently, 
when any covalency is involved, a pure crystal field theory can not explain the 
experimental data very well. 

The advantage of the ligand field theory is the mixing between the electrons of the 
central ion and the ligands. This feature of mixed ionic-covalency successfully explains 
the coloring phenomena for most situations involving transition elements. In this 
theory, a ligand presents a negatively charged, nonspherical, partially covalent bonded, 
distorted coordination complex towards the positive central transition-metal ion. 



209 



For example, in a free ion of a transition metal the five equivalent d orbitals are 
depicted spatially as shown in Figure 6-1. The energy level diagram of the five orbitals 
in a free transition metal ion is also illustrated in Figure 6-2(a). The electrons can be 
found with equal probability in any of these five orbitals (Figs. 6-1, 6-2(a)). When 
this positive transition metal ion with partly filled d-orbitals is placed at the center of a 
regular (undistorted crystal field) octahedron of ligands, represented as point negative 
charges, the configuration is as shown in Figure 6-3. An interaction of the d orbitals of 
the central ion with the six ligands in the octahedral field is expected. The two lobes of 
the d z 2 orbital point exactly at the two ligands in the +Z and -Z directions; similarly, 
the four lobes of the d x 2. y 2 orbital point exactly at the four ligands in the plus and 
minus directions of the X and Y axes. The electrostatic interaction of these two d shells 
with the negative ligands at the corners of the octahedron is repulsive, consequently, 
there is a splitting and raising of the energy of these two of the five d energy levels of the 
system. This leads to the two upper e g levels for the octahedral ligand field configuration 

as shown in Figure 6-2(b). 

The remaining three sets of d orbitals of Figure 6-1, the d X y, d y2 and d 2X 

orbitals, have orientations which protrude halfway between the ligands as shown in 
Figure 6-4. Because there is no repulsive interactions with the ligands these orbitals 
will have lower energy levels than the e g set. Thus, the d xy , d yz and d 2x orbitals are 
shown as the three equal energy t2g levels in Figure 6-2(b). 

The energy difference between the e g and t2g ievels called the crystal field 
splitting, is designated A . Since the overaSJ energy does not change, the upward and 
downward movements are inversely proportional to the number of equal energy levels; 
e.g. e the degeneracy. Thus in Figure 6-2(b), the triply degenerate lower energy t2g 
level moves down 0.4 A , while the upper doubly degenerate e g level moves up 0.6 A 
compared to the unsplit free ion levels (Figure 6-2(b)). 



210 




X 



Figure 6-1 Electron distribution shapes of the five equivalent d obritals. 



211 



Ligand configurations: 



d x 2. y 2 



(a) five unspilt d orbitals in a free ion. 

(b) five splitf&d d orbitals in an octahedral field. 

(c) five splitted-d orbitals in a tetrahedral field. 

(d) five splitted d orbitals in a tetragonally 

distorted octahedral field, 

(e) same as (d) but relatively strong distorted. 

(f) five splifted d orbitals in a square planar 

ligand field. 



i 
i 



d x 2_ y 2 , 



d x 2- y 2 / 



e g : d x 2. y 2 d z 2 



rT\ 



d X y dy Z d ZX 



*t fc.6A t ' / 

Li/ 

d x 2. y 2 d z 2 



i o 



d^ d yz d zx \ 



0.6^ 



d 7 2 \ 



J xy,' 



. >•'■ 



\-U-(. 



^g^xydyzdzx \ 



**A[ = 4/9 4, 



d,2\ 



\ 



dxy_ 



I 

\ 



d 2X dy Z % * % 



d zx d yz 



\ d zx d yz 



d z 2 



(C) 



(a) 



(b) 



(d) 



(e) 



(f) 



Figure 6-2 Spilitting of the five d orbitals in various types of ligand fields. 



212 



negative charged ligand 




d x 2. y 2 + d z 2 



Figure 6-3 Head on interaction of the d z 2 and d x 2. y 2 orbitals of a central ion 
with six ligands in a octahedral field. 



213 



negative charged ligand 




this presents one of d xv , d vz d zx orbitals 



xy "yz "zx 



Figure 6-4 Less interaction of the d xy d yz d zx orbitals of a centra! ion with six 
iigands in a octahedral field. 



214 



When the transition-metal ion is in tetrahedral symmetry as shown in Figures 
6-5 and 6-6, the situation is reversed. The lobes of the d x 2. y 2 or d z 2 orbitals now lie 
in the direction between the ligands, while the lobes of d X y, d yz and d zx orbitals, though 
not pointing directly towards the ligands, lie closer to them. Thus the t2g (d X y, d yz , d zx ) 
orbitals are destabilized with respect to the e g orbitals. For the same strength ligands, 
the tetrahedral scheme At can be related to the A value of the degenerate orbitals by At 
= 4/9 A , as shown in Figure 6-2(c). 

Practically, octahedral arrangements of the ligands around the transition-metal 
ion are often tetragonally distorted. In such a case the two +Z and -Z (d z 2) ligands in 
Figure 6-3 are gradually moving away from the central transition-metal ion, and new 
energy differences among the d orbitals arise. The d z 2 level will fall and d x 2. y 2 level 
will arise equally at the same time. If the two Z ligands are completely removed, the d z 2 
level becomes the lowest energy level in the resulting square planar ligand arrangement, 
since there are now no energy-raising ligands in that direction and the d x 2. y 2 becomes 
the highest energy level, as shown in Figure 6-2(f). For a sequare planar ligand field 
the location of the d yz and d zx levels will fall and that of the d xy level must rise two 
times as much. The frequently observed tetragonally distorted octahedral arrangements 
are shown in Figure 6-2(d) and (e). Consequently, the kind of energy level arrangement 
formed by the ligand fields depends on three crucial properties: (1) the orbitals of the 
central transition-metal ion, (2) the surrounding arrangement of ligand fields, and (3) 
the strength of the ligand fields. 

For ligand fields, in dealing with individual orbitals of an atom, lower case 
notations such as a-| g , b, e-| g , t2g are used. Either a or b indicates a nondegenerate orbital 
with a presenting a wave function which is symmetric with respect to the rotation axis, 
whereas b represents a wave function which is antisymmetric and changes sign during 
rotation. The e and t orbitals are symmetrically doubly and triply degenerate. The energy 
levels in e or t orbitals are equal. 



215 



negative charged ligand 




d x 2. y 2 + d z 2 



Figure 6-5 Interaction of the d z 2 and d x 2. y 2 orbitais of a central ion with four 
ligands in a tetrahedrai field. 



216 



negative charged ligand 




this presents one of d xv , d vz d^ orbitals 



X y> "yz "ZX 



Figure 6-6 Interaction of one of the d xy d yz d zx orbitals of a central ion with four 
ligands in a tetrahedral field. 



217 



Use of the subscript g designates the presence of a change in sign of the wave 
function on inversion through a center of symmetry. A subscript 1 refers to the 
presence of mirror planes parallel to the symmetry axis and a subscript 2 refers to 
mirror planes normal to this axis. Upper case designations such as 2 A2g, 1 Bi, 2 Eg, 2 T2g 
are generally used to represent the energy levels in the atom, ion, or molecule, with the 
prefix superscript as the (2S + 1) multiplicity. 

The energy states that can accommodate undisturbed or excited electrons in free 
transition-metal ions having incomplete d orbitals (d 1 to d 9 ), based on Russell- 
Saunders coupling [see p. 381-408 in ref. 102], are named to be S, P, D, F, G, H and I 
corresponding to the quantum number L equal to 0, 1 , 2, 3, 4, 5, and 6. These states are 
listed in Table 6-1 for various transition metals. 

In a ligand field the tetrahedral d 9 or d 4 configuration can be viewed as containing 
one hole; i.e., one electron missing from a full d or half full d shell. This configuration 
provides a strong analogy with one electron added to an octahedral empty d shell (d 1 ) or 
a half filled d shell (d 6 ) and, conversely, so do the octahedral d 9 or d 4 and the 
tetrahedraf d 1 or d 6 configurations, as shown in Figure 6-7, except that the highest 
rather than the lowest split orbital is being occupied. The same applies to all other 
configurations. For example, the tetrahedral d 2 or d 7 configuration has the same 
sequence of levels as the octahedral d 8 or d 3 and, conversely, so do the octahedral d 2 or 
d 7 and the tetrahedral d 8 or d 3 configurations. These similarities are shown in Figure 
6-8. Consequently, the splitting scheme of d n (octahedral) configurations is equivalent 
to that of d( 10 " n ) (tetrahedral) or vice versa. The # and d 10 configurations 
corresponding to completely empty or completely full d orbitals cannot show color 
directly derived from d electronic transitions. 

An s 1 orbital is completely symmetrical and hence is unaffected by ligands in an 
octahedral field such as 2 S or A-ig. The p 1 orbitals are not split by octahedral fields such 
as 2 P or 2 T-| g since all interact equally as illustrated in Figures 6-9 and 6-10. In an 



218 



Table 6-1 
Energy levels for transition-metal free ions 

Configuration Example The lowest energy levels 

in a free ion 

3d 1 m 3d 9 3d 1 (Ti 3+ ) - 3d 9 (Cu 2+ ) 2 D 

3d 2 =3d 8 3d 2 (V 3+ } - 3d 8 (Ni 2+ ) 3 F< 1 D< 3 P< 1 G< 1 S 

3d 3 =3d 7 3d 3 (Cr 3 *, V 2 +) 4 F< 4 P< 2q <2 H< 2 P 

- 3d 7 (Fe + , Co 2 *) 

3d 4 =3d 6 3d 4 (Cr 2+ , Mn 3+ ) 5 D< 3 H < 3 P < 3 F< 3 G < 1 I 

m 3d 6 (Mn+ , Fe 2+ , Co 3+ ) 

3d 5 3d 5 (Cr+ , Mn 2+ , Fe 3+ ) 6 S < 4 G < 4 P < 4 D < 2 I < 4 F 



219 



T 2a (D) 




T 2a (H) 



E 1 9 (D) T 1g (P) 



a: (hole) d 4 , d 9 In octahedral field 
(electron) d 1 , d s in tetrahedral field 



b: (electron) d 1 , d 6 in octahedral field 
(hole) d 4 , d 9 in tetrahedral field 



Figure 6-7 Three lowest energy levels for d 1 , d 6 , d 4 , d 9 splitting configurations in 
octahedral and tetrahedral fields. 



220 



strength increasing directions 
,9tJyi9.tXP. e .?.9tJLQ? n .^.t)?JSl.. 



T 2a (F) 




T 1 9 ( F ) 



T 2g (D) 



b: (2 holes) d 3 , d 8 in octahedral field 
(2 electrons) d 2 , d 7 in tetrahedral field 



a: (2 electrons) d 2 , d 7 in octahedral field 
(2 holes) d 3 , d 8 in tetrahedral field 



Figure 6-8 Two lowest energy levels for d 2 , d 3 , d 7 , d 8 splitting configurations in 
octahedral and tetrahedral fields. 



221 



negative charged ligand 




symmetrical 

in octahedral field 



Figure 6-9 All interaction between figands and 4S 1 are equal, 
therefore, no splitting results. 



222 



negative charged ligand 




equal P orbital distance 
in X, Y, Z directions 



Figure 6-10 All interaction between ligands and 4P 1 are equal, 
therefore, no splitting results. 



223 



octahedral field the d 1 and d 9 ( 2 D) orbitals, as discussed before, split into T2g (d xy , 
dyz. d zx ) and E g (d x 2. y 2 or d z 2) levels. The f 1 orbitals are split into three levels in an 
octahedral field: a 2 T-| g level at 1/3 A below, a 2 T2g level at 1/9 A above, and a 2 A2g 
level 2/3 A above the prespiitted F orbital as shown in Figure 6-1 1(a). 

The two split low-energy states 3 F and 3 P from either the d 2 or the d 8 
configuration behave in an octahedral field exactly as the F and P states arising from the 
1 f and 1 p as discussed above. Consequently, the 3 F state is split into 3 T-|g(F), 3 T2g(F) 
and 3 A2g(F) states and the unsplit 3 P becomes the 3 Ti g (P) state. The d 3 or d 7 state has 
4 P and 4 F orbitals. Under a ligand field the 4 F splits into 4 Ti g (F), 4 T2g(F) and 4 A2 g (F) 
states and the unsplit 4 P becomes the 4 T-j(P) state, as shown in Figures 6-8 and 6- 
11(b). The d 4 and d 6 configurations have a low-energy state 5 D which splits into 5 T2 g 
(dxy. d yz , d zx ) and 5 E g (d x 2. y 2 or d z 2) in an octahedral field (Figures 6-7 and 6- 
11(c)). The d 5 state has an unsplit 6 S or 6 Ai level in the octahedral field. 

As mentioned above, ligand field theory describes the bonding occurring between 
center transition-metal ion and the ligands. Molecular orbital theory, which develop the 
combination of the atomic orbitals of the atoms to form the molecule, is used to explain 
this bonding phenomenon. 

The condition for two atoms to form: (1) a bonding molecular orbital (vb), (2) a 
nonbonding molecular orbital, or (3) an antibonding molecular orbital (y a ) depends on 
S, the wave function overlap integral J\|/ a ybdt of the probability equation (vb 2 dt = J 
YA^t + J yB 2 dt + J vAVBdt, where ya and yB are the wave functions of atoms A and B) 
for finding an electron within the space. Bonding takes place only when the value of S is 
positive (S > 0) and the bonding strength (energy) is proportional to the extent of the 
overlap of the atomic orbitals. The bonding energy level is reduced relative to the level 
of the free atoms by the same amount as the energy is increased for the antibonding level. 

In the case of 3d transition-metal ions in an octahedral field, the d x 2. y 2 and d 2 2 
configurations are in the direction of the ligands. This results in a positive overlap and a 



224 



(a) octahedral field 
d 2 (V3+) 
d 7 (Co 2+ ) 



r t 18 (p) 




- ± T 1g (F) 



(b) tetrahedral field 
d 3 (V 2+ , Cr 3 *) 
d 8 (Ni 2 +) 




A 2g (F) - 



(c) 



3/54 
D / 




octahedral field: 
d 1 (Ti 3+ ) 
d 6 (Fe 2+ Co 3+ ) 



2/5A 



3/5*. 




D 



tetrahedral field: 
d 4 (Cr 2+ , Mn 3 *) 
d 9 (Cu 2+ ) 



Figure 6-1 1 The splitting of d orbitals (a), (b) for P, F states, 

(c) for D state in octahedral and tetrahedral ligand field. 



225 



reduction (e g , bonding) and an increase (e* g , antibonding) in energy levels. The t2 g 
(d X y, d yz , d zx ) levels are located between the ligands, and no overlap is observed. As a 
result, the energy levels of t2g remain unchanged in the presence of the ligand field as 
shown in Figure 6-1 2(c). The 4s orbital of transition-metal ions has a (aig) spherical 
symmetry and a corresponding ligand group orbital that is composed of sigma bonds 
which are cylindrically symmetrical about the internuclear axis (Figure 6-13). The 4p 
(t-iu bonding and t*i u antibonding) orbitals with the related ligand group orbitals are 
shown in Figure 6-14. 

The 12 electrons from the ligand group orbitals are perfectly paired into three 
lowest energy molecular orbitals which are ai g (1), t-| U (3) and e g (2), as shown by the 
heavy arrows in Figure 6-1 2(c). Consequently, the d electrons from the transition- 
metal ions have to fill the t2g(3) levels first. If there are any remaining electrons then 
the e* g (2) levels available. If the energy gap, A, is greater than kT, low-spin 
configurations will be formed. The gap energy , A, is generally in the visible range of 1 
eV (NIR) to 3 eV (UV) energy. If electromagnetic radiation has the same amount of 
energy as the gap energy, i.e. A = fro (photon energy), then the electronic transitions 
from t_2 g (3) to e* g (2) levels takes place. 

These theories of ligand field and molecular orbital transitions are the basis for 
interpreting the optical spectra of Co 2+ ion doped, Ni 2+ ion doped and Cu 2+ ion doped 
silica gels. The possible energy level transitions for the Co 2+ ion in both d 7 octahedral 
and tetrahedral symmetries are analyzed by means of Figures 6-8(a) and (b). The 
octahedral Co 2+ configuration is predicted to have three spin-allowed transitions: 

(1) 4 Ti(F) -> 4 T 2 (F), 

(2) 4 T1 (F) --> 4 Ti(P), 

(3) 4 Ti(F) -> 4 A 2 (F). 

The tetrahedral Co 2+ configuration is expected to have three major transitions: 
(1) 4 A 2 (F) --> 4 T 2 (F), 



226 



4p 
empty 



empty 
4s" 



partially 
3d filled 



(a) free ion 
orbitals 




(b) central ion 
orbitals 



completely filled 
ligand group orbitals 



#4WM 



(d) ligand orbitals 



tiuO) 



(c) molecular orbitals 



Figure 6-12 Molecular orbital splitting levels for a d orbital ion in an octahedral 
environment with ligands having only c bonds. 



227 



(a) 




q bonding ligand group orbitals 



(b) 




a antibonding ligand group orbitals 



Figure 6-13 Ligand group orbital and central matching atomic orbitals 
of the bonding symmestry. 



228 



(a) 




a bonding ligand group orbitals 



(b) 



central ion 4p orbital 




ct antibonding ligand group orbitals 



Figure 6-14 Ligand group orbital of only a bonds and matching atomic orbitals 
to form molecular orbitals. 



229 



(2) 4 A1 (F) --> 4 T i(F), 

(3) 4 A 2 (F) --> 4 T!(P). 

The energy level diagram for the Ni 2+ ion in d 8 tetrahedral and octahedral 
symmetries is also depicted in Figures 6-8(a) and (b). The tetrahedral Ni 2+ 
configuration has three spin-allowed transitions: 

(1) 3 T1 (F) -> 3 T2 (F), 

(2) 3 T1 (F) --> 3A 2 (F), 

(3) 3 T1 (F) --> 3 Ti(P). 

The octahedral Ni 2+ symmetry results in three major transitions: 

(1) 3 A2 (F) -> 3 T2 (F), 

(2) 3 A1 (F) -> 3 T1 (F), 

(3) 3 A 2 (F) --> 3 T1 (P). 

The Cu 2+ ion has a 3d 9 configuration, an inverted d 1 configuration, as shown in 
Figure 6-7. The major transition is attributed to 2 E --> 2 T 2 , as is also shown in Figure 
6-7(a). 

Experimental Procedure 
Seven steps are generally used in making the monolithic silica gels and glasses 
containing transition-metal elements: (1) mixing, (2) casting, (3) gelation, (4) aging, 
(5) drying, (6) impregnation and (7) densification described in Example Two in 
Chapter 2. In the mixing stage, it is necessary to select a suitable drying control 
chemical additives, such as formamlde, glycerol, nitric acid, or an organic acid, in order 
to make monoliths rapidly without: (1) precipitation, (2) formation of an 
inhomogeneous gel, or (3) crystallization. By use of nitric acid in this system it was 
possible to produce non-crystalline homogeneous optical silica gels and glasses. To our 
knowledge, monolithic gels containing the transition and rare earth elements mentioned 
in Chapter 2 have not previously been described. 



230 



The examples used for this investigation were Co", Ni" and Cu" colored silica 
monoliths. The first step involved mixing 60 cc (1N) nitric acid DCCA with 340 cc of 
distilled water for 5 minutes at room temperature, followed by adding to the nitric acid 
water solution 200cc of TMOS with mixing at 85°C for no more than 60 minutes. This 
well mixed sol was then cast into a polystyrene mode (20 mm H x 100 mm D, a disk 
shape) at room temperature. Gelation occurred in the mold at 55°C in about 115 
minutes, followed by aging at 55°C for 10 hours and followed by aging at 80°C for 15 
hours. The aged silica gel was taken from the molds and dried with a controlled 
evaporation rate, as described in Section II of Chapter 2. The drying was initially at 
70°C with the temperature gradually increasing to 160°C during a 90 hour period. 
Before impregnation, the gel was stabilized to 800°C at 10°C/hour to increase the 
strength and density and make it possible to perform a nondestructive doping process. 
The stabilized gel was then immersed into a 0.25 gram-percent Co" nitrate or a 0.30 
gram-percent Ni" nitrate or an one gram-percent Cu" nitrate water solution for 24 
hours. The solution doping followed by drying at 160°C for 12 hours to remove the pore 
solvent. Subsequent thermal treatments to 850°C and 900°C were done in ambient air. 

The transmission spectra of the 160°C Co" and Cu" doped silica gel glasses and 
the 850°C, 900°C Co" doped silica gel glasses were obtained in the visible range from 
200 nm to 900 nm using a Perkin-Elmer UV-VIS spectrophotometer model 552. The 
transmission spectra of the 160°C Ni" doped silica gel glass was performed in the UV- 
VIS-NIR range from 200 nm to 1300 nm using a Perkin-Elmer Lambda 9 UV-VIS-NIR 
spectrophotometer. 

Results and Discussions 
The silica gel samples containing 0.25% Co were heated to certain temperatures. 
The color of the 160°C Co" gel is reddish pink. The color of the 850°C sample is deep 
blue, and the 900°C sample has a greenish black color. The UV-Visible spectra 



231 



characteristic of these three Co"-silica gel samples are shown in Figure 6-15. There is 
a totally different absorption curve for the 160°C pink sample than for the 850°C blue 
sample and the 900°C green sample. Since the color of transition ions such as cobalt in 
silicate glasses depends primarily on the outer d valence orbitals, it means that the color 
and absorption spectra depends on the oxidation state and coordination number of the ion. 
The temperature sensitivity of the Co"-silica gel absorption spectra indicates a shift in 
oxidation state and coordination number (CN). The Sow-temperature gel shows evidence 
of a sixfold CN similar to that reported for Co" in metaphosphate glasses [117] and 10 
mol% Na20-borate glass [see p. 241 in ref. 112], as shown in Figure 6-16. Thus, it is 
reasonable to assume that the Co" ion in the silica gel in octahedral symmetry. 

The major absorption band of the Co" ion is due to the 4 Ti(F) to 4 Ti(P) 
transition (see Figure 6-16). The high energy shoulder at 470 nm is a consequence of 
spin-orbit couping in the 4 T-|(P) state [9]. The 4 T-|(F) to 4 T2(F) transition occurs in 
the infrared region around 1250 nm and does not contribute to color formation. The 
4 T-|(F) to 4 A2(F) transition is expected to be at 555 nm. However, this transition is 
very weak because it involves the forbidden two-electron jump [118]. This weakness 
combined with the closeness of the major 4 Tt(F) to 4 Ti(P) transition makes the 4 T-|(F) 
to 4 A2(F) transition unresolved. 

in contrast, the high-temperature (850°C and 900°C) Co" doped gels appear to 
have a CN of 4. This fourfold coordination is more equivalent Jo that of a standard 
vitreous silicate glass [119] (see Figure 6-17), that is, 
Co 5l 6 (pink) — £— > Co l! 4 (blue). 

The main absorption band in 550 nm to 700 nm range of this tetrahedral Co" ion 
is due to the 4 A2(F) to 4 T-|(P) transition. As shown in Figure 6-17, the Co" doped high- 
temperature gel shows evidence of a fourfold CN similiar to that for Co" ion in fused 



232 



1.0 



0.8 



0.6 



a 
o 



w 

a 

2 0.4 
o 

a. 
o 



0.2 



0.0 




200 300 400 500 600 

wavelength (nm) 



700 



800 



900 



Figure 6-15 Spectra of three Co !l -doped silica gel samples at 160°C, 850°C, and 900°C. 



233 




300 400 



500 600 

wavelength (nm) 



Figure 6-16 Spectra of 160°C Co"-doped silica gel samples and some Co"-doped 
melted glasses. 



234 



1.0 



0.8 



0.6 



Q 
O 



w 

c 

3 0.4 

o 

E. 
o 



0.2 



0.0 




200 300 400 500 600 

wavelength (nm) 



700 



800 



900 



Figure 6-17 Spectra of 850°C, 900°C Co ss -siiica gel samples and two Co n -doped 
melted glasses. 



235 



silica and a binary 30 mol% Na20-borate glass. 

The splitting of the 4 A2(F) to 4 T-|(P) band is caused by spin-orbit coupling 
which splits the 4 T-|(P) states and allows the transitions to the neighboring doublet 
states to gain in intensity [see p. 241-242 in ref. 112]. The two other transitions, 
4 A2(F) to 4 T2(F) and 4 A2(F) to 4 T-|(F) which take place in the infrared region 
contribute no color chromophores. In this study, none of the spectra for the Co" doped 
silica gels is identical to the silicate melt glass spectrum in detail. This indicates that the 
ligand field strength (A) may be varied by the thermal history of the gels. 

The spectrum of a 160°C Ni" doped silica gel is similar to that of a 16.2 wt.% 
melt K20-borate glass containing Ni ion. It is also similar to that of a [Ni(H20)6] 2+ 
octahedral complex in water [see p. 242-243 in ref. 112], as shown in Figure 6-18. 
The absorption band at 700 nm of Ni 2+ in an octahedral complex is assigned to the 
3 Ai(F) --> 3 Ti(F) transition, and the one at about 400 nm is assigned to the 3 A2(F) - 
-> 3 Ti(P) transition. Another band corresponding to a 3 A2(F) --> 3 T2(F) transition is 
observed in the infrared region at about 1180 nm. In this study, the spectra of these 
three samples are almost the same except for the difference in absorption intensity. The 
similarity in absorption bands of the three curves indicates that the same ligand field 
strength acts on Ni 2+ ion in these three samples. 

The absorption spectrum of Cu" in a 160°C gel and three binary sodium-borate 
Cu" melt glasses [120] are shown in Figure 6-19. All the absorption spectra consist of 
a broad band with a maximum at about 780 nm. This absorption is attributed to the 
transition from 2 E levels to 2 T2 levels. The band is asymmetric and departs from 
Gaussian symmetry since the 2 T2 levels are split by a distorted low symmetry ligand 
field component. No significant band shift and shape change is present in spite of the 
variations of the surrounding ligand chemical composition of the ligands. 



236 



1.0 



0.8 



Q 

o 



■a 

s 

'a. 

o 



0.6 



0.4 



0.2 



0.0 




350 



550 



750 950 
wavelength (nm) 



1150 



1350 



Figure 6-18 Absorption spectra of a Ni"-dopec! siiica gel sample, a Ni" water 
solution and a Ni"-doped 16.2 mol% K 2 0-B 2 3 glass. 



237 



1.0 



0.8 



0.6 



O 
O 0.4 



en 

C 

■o 



CO 

o 



o 



0.2 



0.0 




350 



550 



750 950 

wavelength (nm) 



1150 



1350 



Figure 6-19 Absorption spectra of Cu !L doped silica gel sample and three 
Cu"-doped sodium-borate glasses. 



238 



Conclusions 
Thermal history can alter the chemical environment and ligand-field around a 
transition metal ion in a silica gel and have a marked effect on its optical absorption 
characteristics and hence on the color produced. These results show that it is possible to 
take advantage of low temperature so!-gel-glass techniques to manufacture various 
optical filters using a silica matrix. The absorption spectra can be shifted by controlling 
the thermal history of silica gels containing transition elements. The optical components 
produced will have the unique physical properties of silica as discussed in previous 
chapters, that is, low thermal expansion coefficients, extraordinarily high chemical 
durability, and superb thermal shock resistance. In addition, depending upon the extent 
of densification reached during thermal processing the density and index of refraction of 
the optical component can be varied over wide ranges. The flexibility of the sol-gel 
technique offers such new exciting processing methods in the production of a variety of 
optical components. The silica gel glasses chemically doped with transition-metal ions 
(Co, Ni, Cu ions) discussed in this chapter have demonstrated the possibilities of 
producing a variety of products including strategic high-tech optical glasses with 
specific wavelength filtration capabilities, high-tech commercial sun glasses, and 
tuneable laser glasses. 



CHAPTER? 
CONCLUSIONS AND RECOMMENDATIONS 



Sol-gel processing offers a new manufacturing method for high technology 
ceramics and glasses since it allows structural manipulation down to the molecular scale 
in the nanometer range. Thus, ultrahigh purity and extreme molecular homogeneity of a 
material may be achieved. 

Two major chemical reactions, hydrolysis and polymerization, are involved in sol- 
gel ultrastructure processing. Hydrolysis enables the organometallic chemical 
precursor to react to form a monomer on the atomic scale, which is composed of a 
positive metallic ion surrounded by an anionic complex (e.g., Si +4 (OH _ )4. Ti +4 (OH - )4, 
AI +3 (OH")3, Si +4 (CH3")4 etc.). This step is followed by a polymerization based 
growth process which links the monomers together. 

In recent years special optical applications require silica components that meet 
very stringent requirements. Sol-ge! processing applied to silica offers the potential for 
producing a new generation of silica glasses to meet these requirements for optical and 
electro-optical applications. The quality of gel-silica glasses expected to meet these 
stringent requirements are {1} very high purity, (2) extremely low optical signal 
loss, (3) very high chemically homogeneous doping, (4) very high optical homogeneity. 
These features make gel-silicas able to upgrade the optical performance in a wide range 
of precise optical apparatus including lenses, mirrors, waveguides, optical fibers, 
integrated optoelectronics, and host materials for filters, lasers, and non-linear optical 
elements or compounds. Therefore, achieving a chemically optimized sol-gel processing 
for silica optical monoliths was the focus of this study. 



239 



240 

The first major difficulty faced in producing large monolithic gel glass for optical 
components was cracking during drying. In this study the problem was overcome by use 
of drying control chemical additives (DCCA) and a special designed drying chamber 
described in Chapter 2. 

A fibrillar structure of a silica gel can be formed in the initial preparation of 
silicic solution. Once the ratio of silica precursor, water and DCCA is fixed the fibrillar 
structure of the gel is determined. Further evolution of the structure is a function of 
time and temperature. A relatively strong gel was made using an acidic DCCA which 
enabled the gel to endure the catastrophic capillary forces developed inside the gel during 
drying. In addition, the introduction of special drying chamber for ambient atmosphere 
control also reduced the capillary stress significantly. Monolithic dried (physical water 
free) gels as large as 10 cm x 8 cm x 2.0 cm {up to the capability of the experimental 
facility) were routinely produced. The first goal was achieved. 

The dried gel monoliths were partially densified in an ambient air furnace up to 
860°C. The characterization of these partially densified silica gels was performed by use 
of (1) structural information tests (x-ray diffraction, BET), (2) Optical information 
tests (refractive index, FTIR, UV-VIS-NIR), (3) thermal information tests (DSC, DTA, 
TGA, TMA), (4) mechanical information tests (flexural strength, compressive strength, 
microhardness, toughness, density). 

The results of structural information tests showed that the gels were an amorphous 
phase with high a volume, and tremendous surface area of uniform open pores 
throughout the entire body with chemisorbed hydroxyl groups being a function of 

temperature. 

The conclusion of optical information test proves that the index of refraction is a 
function of sintering temperature and the index has a linear relationship with density as 
predicted by the Lorentz-Lorenz equation. The -OH absorption bands and UV cut-off is 
also a function of sintering temperature. 



241 

The thermal information tests indicate that the decomposition and evaporation 
weight losses due to loss of residual organic compounds and water take place below 
450°C, Dimensional shrinkage occurs throughout the entire heating program. 

The results of mechanical information tests show that the compressive strength, 
maximum strain to fracture, fiexura! strength, Young's modulus, and microhardness are 
linearly proportional to the gel density. These values show a tendency to approach the 
values of fused silica as the gel densification temperature increases. The K|c/density data 
obtained show the 160°C gel has a greater toughness than the value for fused silica 
proving the fibrillar gel structure can absorb higher impact energy than fused silica 
before cracking. 

The second major difficulty to be overcome in producing large monolithic gel glass 
for optical components was the surface silanol groups inside the porous silica gel which 
terminated the -O-Si-O- bridging bonds and degenerated the optical performance 
significantly. A dehydration thermal treatment using carbon tetrachloride was 
accomplished. Monolithic samples of fully dehydrated and densified monolithic pure 
silica gel-glass were routinely reproduced. 

The tests of the dehydrated densified gel-silica monoliths yielded very important 
results. A very high optical transmission was achieved throughout the entire spectral 
range between 165 nm and 4400 nm in the VUV-UV-VIS-NIR spectra. The vuv cut-off 
wavelength was at 162 nm. These results were equivalent to the very best Type IV 
commercial silicas. The CTE data of the gel-silica had almost three times lower (2.0 x 
10" 7 ) values than that of Corning 7940 pure silica (5.5 x 10" 7 ). Thus, the monolithic 
gel-silica glass greatly improved the optica! transmission and significantly lowered 
thermal expansion of traditional Type l-!V silica glasses. Equivalent or superior levels 
of homogeneity, strain, bubbles and striae were also achieved. The second goal of this 
study was reached. 



242 

Applications are nearly unlimited for use of the silica sol-gel technology developed 
herein. The possibility of using the porous gel monoliths for second phase doping was the 
third goal pursued. Monolithic partially densified optical silica-gel filters impregnated 
with transition-metal ions (i.e., Cu +2 , Ni +2 , and Co +2 ions) were successfully made. 
Color changes of these transition-metal ion doped gels that resulted from different 
densification temperature were interpreted using ligand field and molecular orbital 
theories. The third goal in this study was achieved. 

The highest quality of pure silica made in the world today is that of optical fibers 
fabricated by vapor phase plasma reaction of ultrapure oxygen with ultrapure silicon 
tetrachloride (Type IV silica). This process results in fibers of ultralow loss (about 1.0 
dB/Km to 5.0 dB/Km) in the 900 nm to 1300 nm range. It is already shown in this 
study that the fully dehydrated, completely densified, gel-glass monoliths are of such a 
quality as to compare with the best Type IV optical silica fibers. However, the 
temperature of densification has been reduced to 1150°C. The sol-gel silica process has 
the additional advantage that net shape casting of optical components is very simple, also 
localized densification can be achieved yielding a new approach for producing waveguides 
in a pure silica matrix (i.e., integrated optics) 

The vuv cut-off at 162 nm indicates that the gel-glass is not yet an ideal silica 
glass. Test data from neutron activation analysis showed that the impurity levels in the 
first generation silica gel-glass were reduced to several ppb or even better than that of 
Type IV silica. However, a significant chlorine content, at a value 0.1 wt%, was present 
which caused a serious problem of the increasing of index of refraction and foaming of a 
sintered gel-glass above 1300°C. In addition, chlorine terminates the bridging oxygen 
bond, limits the theoretical silica performance, and create a possible optical absorption 
center in high energy electromagnetic radiation fields (e.g., x-ray, gamma ray). 
Elimination of the chlorine impurity is the most important subject for improvements in 
sol-gel processing if the ultimate performance in silica glass is required. 



243 

There is no difficulty for the sol-gel process to prepare an extremely intimate and 
chemically homogeneous sol to form a molecuiarly uniform gel. The optical homogeneity 
problem due to localized density and chlorine fluctuations could be improved by 
developing further optimization of the thermal dehydration densification process. This 
will require modification of the atmosphere control system of the furnace. 

Use of the chemical doping technique for porous gel-silica could lead to produce a 
new category of multicomponent glasses. For instance, a colorful and fascinating world in 
the transition-metal and rare earth elements doped gel-glasses is waiting for further 
exploration. Oxidation states and ligand fields can be stabilized within the gel-silica 
matrix that are not possible using traditional high temperature melt derived glasses. 



REFERENCES 



1 . B. Jirgensons and M. E. Straumanis, Colloid Chemistry . Macmillan Co., New 
York, 1962. 

2. Ralph K. Her, The Colloid Chemistry of Silica and Silicates . Cornell University 
Press, Ithaca, New York, 1955. 

3. P. J. Flory, Gels. A Introduction Lecture, Faraday Discussions of the Chemical 
Society, Vol. 57, 1974, p. 7-18. 

4. Ralph K. Her, The Chemistry of Silica . John Wiley & Sons, Inc., New York, 
1979. 

5. R. H. Doremus, Chemical Durability of Glass, in Treatise on Materials Science 
and Technology Volume 17, Glass II . Miknoru Tomozawa and Robert H. Doremus, 
eds., Academic Press, Inc., New York, 1979, p. 41-69. 

6. C. J. Brinker, K. D. Keefer, D. W. Schaefer and C. S. Ashley, Sol-Gel Transition in 
Simple Silicate, Journal of Non-Crystalline Solids, Vol. 48, 1982, p. 47-64. 

7. L C. Klein, Sol-Gel Glass Technology. A Review, Glass Ind., 1981, p. 14-16. 

8. J. D. Mackenzie, Fusion of Quartz and Cristobalite, Journal of the American 
Ceramic Society, Vol. 43, 1960, p. 615-620. 

9. Martin Grayson, ed., Encyclopedia of Glass, Ceramics, and Cement . John Wiley 
& Sons, Inc., New York, 1985, p. 837-845. 

10. N. J. Kreidl, Inorganic Glass-Forming Systems, Part I: Vitreous Silica, in Glass: 
Science and Technology Vol. 1: Glass-Forming Systems . Academic Press, Inc., 
New York, 1983, p.107-121. 

11. J. D. Mackenzie, Glasses from Melts and Glasses from Gels, A Comparsion, 
Journal of Non-Crystalline Solids, Vol. 48, 1982, p. 1-10. 

12. L L Hench, Concepts of Ultrastructure Processing, in Ultrastructu re Processing 
of Ceramics. Glasses and Composites . L L. Hench and D. R. Ulrich, eds., John 
Wiley & Sons, Inc., New York, 1984, p. 3-5. 

13. Sumio Sakka, Gel Method for Making Glass, in Treatise on Materials Science and 
Technology Volume 22. Glass III . Miknoru Tomozawa and Robert H. Doremus, 
eds., Academic Press, Inc., New York, 1982, p. 129-169. 

14. I. Artaki, M. Bradley, T. Zerda, Jiri Jonas, G. Orcel, and L L Hench, NMR, 
Raman Study of the Effect of Formamide on the Sol-Gel Process, in Science of 
Ceramic Chemical Processing . L L. Hench and D. R. Ulrich, eds. John Wiley & 
Sons, Inc., New York, 1986, pp. 73-80. 



244 



245 



15. L L Hench, Use of Drying Control Chemical Additives (DCCAs) in Controlling 
Sol-Gel Processing, in Science of Ceramic Chemical Processing . L L. Hench and 
D. R. Ulrich, eds., John Wiley & Sons, Inc., New York, 1986, p. 52-63. 

16. G. Orcel and L L Hench, Effect of the Use of a Drying-Control Chemical Additive 
(DCCA) on the Crystallization and Thermal Behavior of Soda Silicate and Soda 
Borosilicate, Proceedings of the 8th Annual Conference on Composites and 
Advanced Ceramic Materials, Cocoa Beach, Florida, January 15-18, 1984. 

17. S. Wallace and L L Hench, Metal Organic Derived 20L Gel Monoliths, 
Proceedings of the 8th Annual Conference on Composites and Advanced Ceramic 
Materials, Cocoa Beach, Florida, January 15-18, 1984. 

1 8. Donald R. Ulrich, Chemical Science's Impact on Future Glass Research, Ceramic 
Bulletin, Vol. 64, No. 11, 1985, p. 1444-1448. 

19. S. H. Wang and L L Hench, Drying Control Additives for Rapid Production of 
Large Sol-Gel Monoliths Containing Transition and Rare Earth Elements, patent 
pending, Serial No. 704917, 1985. 

20. Gerard Orcel and L. L Hench, Effect of Formamide Additive on the Chemistry of 
Silica Sol-Gels, Journal of Non-Crystalline Solids, Vol. 79, 1986, p. 177-194. 

21. J. Lyklema, The Determination of the IEP and the PZC in Silicic Solution, Faraday 
Discussions of the Chemical Society, No. 52, 1971, p. 318-325. 

22. R. L. Mozzi and B. E. Warren, Structure of Vitreous Silica, Journal of Appl. 
Cryst., Vol. 2, 1969, p.164 -172 . 

23. W. D. Kingery, H. K. Bowen and D. R. Uhlmann, Introduction to Ceramics . 2nd 
ed., John Wiley & Sons, Inc., New York, 1976, p. 95-108. 

24. D. E. Clark, C. G. Pantano and L L Hench, Corrosion of Glass. Books for Industry, 
Div. of Magazines for Industry, New York, 1979. 

25. M. Prassas, J. Phalippou and J. Zarzycki, Sintering of Monolithic Silica 
Aerogels, in Science of Ceramic Chemical Processing . L. L Hench and D. R. 
Ulrich, eds., John Wiley & Sons, Inc., New York, 1986, p. 156-167. 

26. J. Wong and C. A. Angell, Glass Structure bv Spectroscopy . Marcel Dekker, Inc., 
New York, 1976. 

27. Michael L. Hair, Infrared Sp ectroscopy in Surf ace Chemistry . Marcel Dekker, 
Inc., New York, 1967. 

28. E. ML Rabinovich, D. L Wood, D. W. Johnson Jr, D. A. Fleming, S. M. Vincent and 
J. B. MacChesney, Elimination of CI2 and H2O in Gel Glasses, Journal of Non- 
Crystalline Solids, Vol. 82, 1986, p. 42-49. 

29. B. N. Figgis, Introduction to Liganri Fields . John Wiley & Sons, Inc., New York, 
1966. " 



246 



30. L. C. Klein and G. J. Garvey, Monolithic Dried Gels, Journal of Non-Crystalline 
Solids, Vol. 48, 1982, p. 97-104. 

31. M. Decottignies, J. Phalippou and J. Zarzycki, Synthesis of Glasses by Hot- 
Pressing of Gels, Journal of Materials Science, Vol. 13, 1978, p. 2605-2618. 

32. J. Phalippou, M. Prassas and J. Zarzycki, Crystallization of Gels and Glasses 
Made from Hot-Pressed Gels, Journal of Non-Crystalline Solids, Vol. 48, 1982, 
p. 17-30. 

33. R. Roy, Gel Route to Homogeneous Glass Preparation, Journal of American 
Ceramic Society, Vol. 52, 1969, p. 344-345. 

34. B. E. Yoldas, Monolithic Glass Formation by Chemical Polymerization, Journal of 
Materials Science, Vol. 14, 1979, p. 1843-1849. 

35. G. Carturan, V. Gottardi and M. Graziani, Physical and Chemical Evolutions 
Occurring in Glass Formation from Alkoxides of Silicon, Aluminum and Sodium, 
Journal of Non-Crystalline Solids, Vol. 29, 1978, p. 41-47. 

36. M. Yamane, S. Aso, S. Okano and T. Sakaino, Preparation of a Gel from Metal 
Alkoxide and Its Properties as A Precursor of Oxide Glass. Journal of Materials 
Science, Vol. 13, 1978, p. 865-871. 

37. L. L. Hench and Gerard Orcel, Physical-Chemical and Biochemical Factors in 
Silica Sol-Gels, Journal of Non-Crystalline Solids, Vol. 82, 1986, p. 1-10. 

38. Gerard Orcel, The Chemistry of Silica Sol-Gel, Ph. D. Dissertation, University of 
Florida, Gainesville, Florida, 1987. 

39. Z. Z. Vysotskii and D. N. Strazhesko, The Role of Polymerization and 
Depolymerization Reactions of Silicic Acid, etc., in D. N. Strazhesko, ed., 
Adsorption and Adsorbents . John Wiley & Sons, Inc., New York, 1974, p. 55-75. 

40. C. Okkerse, Chapter 5: Porous Silica, in Physical and Chemical Aspects of 
Adsorbents and Catalysts . B. G. Linsen, ed., Academic, New York, 1970, p. 214- 

219. 

41 . S. G. De Bussetti, M. Tschapek, and A. K. Helmy, Calorimetric Determination of 
the Point of Zero Charge, Journal of Electroanalytical Chemistry and Interracial 
Electrochemistry, Vol. 36, 1972, p. 507-511. 

42. Ralph K. Her, Polymerization of Silica Acid: Retarding Effect of Chromate Ion, 
Journal of Physical Chemistry, Vol. 56, 1952, p. 678-679. 

43. Ralph K, Her, Chapter 2: Dissolution and Polymerization of Silica, in Surface and 
Colloid Science . Vol. 6, E. Matijevic, ed., John Wiley & Sons, Inc., New York, 
1973, p. 4-15. 

44. Michael D. Sacks and Rong-Shenq Sheu, Rheological Characterization During the 
Sol-Gel Transition, in Science of Ceramic Che mical Processing. L L Hench and D. 
R. Ulrich, eds. John Wiley & Sons, Inc., New York, 1986, p.100-107. 



247 



45. Paul J. Flory, Condensation Polymerization and Constitution of Condensation 
Polymers, in R. E. Burk and Oliver Grummitt, eds., High Molecular Weight 
Organic Compounds (Frontiers in Chemistry, Vol. VI), Interscience Publishers, 
New York, 1949, p. 211-283. 

46. Paul J. Flory, Fundamental Principles of Condensation Polymerization, Chemical 
Reviews, Vol. 39, 1946, p. 137-197. 

47. Ralph K. Iler, Inorganic Colloids for Forming Ultrastructures, in Science of 
Ceramic Chemical Processing . L L Hench and D. R. Ulrich, eds., John Wiley & 

Sons, Inc., New York, 1986, p. 3-20. 

48. J. Zarzycki, Monolithic Xero- and Aerogels for Gei-G!ass Processes, in 
Ultrastructure Processing of Ceramics, Glasses and Composites . L. L Hench and 
D. R. Ulrich, eds., John Wiley & Sons, Inc., New York, 1984, p. 27- 42. 

49. David R. Gaskell, Introduction to Metallurgical Thermodynamics . 2nd ed. 
McGraw-Hill Book Co., New York, 1981. 

50. J. F. Goodman and S. J. Gregg, The Production of Active Solids by Thermal 
Decomposition, Part X: Heat Treatment of the Xerogels of Silica, Journal of 
the Chemical Society, Vol. 1, 1959, p. 694-698. 

51 . S. Sakka and K. Kamiya, The Sol-Gel Transition in the Hydrolysis of Metal 
Alkoxides in Relation to the Formation of Glass Fibers and Films, Journal of Non- 
Crystalline Solids, Vol. 48, 1982, p. 31-46. 

52. B. E. Yoldas, Effect of Molecular Separation on the Hydrolytic Polycondensation of 
Si(OC2H5)4, Journal of Non-Crystalline Solids, Vol. 82, 1986, p. 11-23. 

53. Michel Prassas and L L Hench, Physical Chemical Factors in Sol-Gel 
Processing, in Ultrastructure Processing of Ceramics. Glasses and Composites. L. 
L Hench and D. R. Ulrich, eds., John Wiley & Sons, Inc., New York, 1984, p. 
100-125. 

54. T. Kawaguchi, H. Hishikura, J. lura, and Y. Kokubu, Monolithic Dried Gels and 
Silica Glass Prepared by the Sol-Gel Process, Journal of Non-Crystalline Solids, 
Vol. 63, 1984, p. 61-69. 

55. S. P. Mukherjee, Sol-Gel Processes in Glass Science and Technology, Journal of 
Non-Crystalline Solids, Vol. 42, 1980, p. 477-488. 

56. Iwao Matsuyama, Kenzo Susa, and Tsuneo Suganuma, Syntheses of High-Purity 
Silica Glass by the Sol-Gel Method, American Ceramic Society Bulletin, Vol. 63, 
No. 11, 1984, p. 1408-1411. 

57. C. J. Brinker, E. P. Roth, D. R. Tallant, and G. W. Scherer, Relationships Between 
Sol to Gel to Glass Conversions: Structure of Gels During Densification, in Science 
of Ceramic Chemical Processing . L. L Hench and D. R. Ulrich, eds., John Wiley & 
Sons, Inc., New York, 1986, p. 37-51. 



248 



58. Gerard Orcel, J. Phalippou, and L L Hench, Structural Changes of Silica 
Xerogels During Low Temperature Dehydration, Journal of Non-Crystalline 
Solids, Vol. 88, 1986, p. 114-130. 

59. T. Izawa and S. Sudo, Optical Fibers: Materials and Fabrication . KTK Scientific 
Publishers, sold by Kluwer Academic Publishers, Norwell, Massachusetts, 
1987, p. 33. 

60. D. C. Havard and R. Wilson, Pore Measurements on the SCi/IUPAC/NPL Meso- 
Porous Silica Surface Area Standard, Journal of Colloid and Interface Science, 
Vol. 57, 1976, p. 276-288. 

61. Clarence L Babcock, Refractive Index and Dispersion, in Silicate Glass 
Technology Methods . John Wiley & Sons, Inc., New York, 1977, p. 87-114. 

62. Du Pont Company, Du Pont 1090 Thermal Analysis System Manual . Du Pont Co., 
Wilmington, Delaware, 1983. 

63. D. G. Holloway, The Physical Properties of Glass . Wykeham Publications Ltd., 
London, 1973, p. 143-149. 

64. G. R. Anstis, P. Chantikul, B. R. Lawn, and D. B. Marshall, A Critical Evaluation of 
Indentation Techniques for Measuring Fracture Toughness: I and II, Direct Crack 
Measurements, Journal of American Ceramic Society, Vol. 64, 1981, p. 533- 
543. 

65. Perkin-Elmer Company, Perkin-Elmer Lambda 9 UV/VIS/NIR Spectrometer 
Manual . Perkin-Elmer Co., West Germany, 1986. 

66. Quantachrome Corporation, Autosorb-6 Manual . Quantachrome Corp., Syosset, 
New York, 1985. 

67. E. P. Barrett, L. G. Joyner and P. P. Halenda, The Determination of Pore Volume 
and Area Distributions in Porous Substances, I: Computations from Nitrogen 
Isotherms.Journal of the American Chemical Society, Vol. 73, 1951, p. 373- 

380. 

68. Jenkins and White, Fundamental of Optics . 3rd ed., McGraw-Hill, New York, 
1957. 

69. ASTM D790M-84, in Annual Book of ASTM Standards . 1986. 

70. ASTM C1 58-80, in Annual Book of ASTM Standards . 1986. 

71. D. B. Keck, R. D. Maurer and P. C. Schultz, On the Ultimate Low Limit of 
Attenuation in Glass Optical Waveguides, Applied Physics Letters, Vol. 22, 1973, 
p. 307-309. 

72. G. H. Sigel, Ultraviolet Spectra of Silica Glasses: A Review of Some Experimental 
Evidence, Journal of Non-Crystalline Solids, Vol.13, 1973/1974, p. 378-398. 



249 



73. J. Phalippou, T. Woignier, and J. Zarzycki, Behavior of Monolithic Silica 
Aerogels at Temperatures Above 1000°C, in Ultrastructure Processing of 
Ceramics. Glasses and Composites . L. L Hench and D. R. Ulrich, eds., John Wiley 
& Sons, Inc., New York, 1984, p. 70-87. 

74. B. D. Cullity, Elements of X-Rav Diffraction . 2nd ed., Addison-Wesley 
Publishing Co., Inc., Reading, Massachusetts, 1978, p. 99-105. 

75. C. A. Mulder, J. G. Van Lierop and G. Frens, Densification of Si02-Xerogels to 
Glass by Ostwald Ripening, Journal of Non-Crystalline Solids, Vol. 82, 1986, p. 
92-96. 

76. Dynasil Corporation of America, Catalog 302-M of Dynasil Synthetic Fused 
Silica, Dynasil Corp. of America, Berlin, New Jersey, 1987. 

77. S. Palmqvist, Occurrence of Crack Formation During Vickers Indentation as a 
Measure of the Toughness of Hard Metals, Arch. Eisenhuttenwes., Vol. 33, No. 6, 
1962, p. 629-633. 

78. F. Orgaz and H. Rawson, Characterization of Various Stages of the Sol-Gel 
Process, Journal of Non-Crystalline Solids, Vol. 82, 1986, p. 57-68. 

79. J. B. Peri and A. L Hensley, Jr., The Surface Structure of Silica Gel, 
Journal of Physical Chemistry, Vol. 72, [8], 1968, p. 2926-2933. 

80. George H. Sigel, Jr., Interaction with Electromagnetic Radiation, in Treatise on 
Materials Science and Technology Volume 12-Glass I . Miknoru Tomozawa and 
Robert H. Doremus, eds., Academic Press, Inc., New York, 1977, p. 14. 

81. M. AN Omar, Elementary Solid State Physics . Addison-Wesley Publishing Co., 
Inc., Reading, Massachusetts, 1975, p. 86-133. 

82. Allen H. Cherin, Fabrication of Optical Fibers, in An Introduction to Optical 
Fibers . McGraw-Hill Book Co., New York, 1983, p. 147-153. 

83. Siemens, Fiber Optic Cables . John Wiley & Sons, Inc., New York, 1987, p. 
32. 

84. Michael L. Hair and William Hertl, Reactions of Chlorosilanes with Silica 
Surfaces, Journal of Physical Chemistry, Vol. 72, 1968, p. 2372-2378. 

85. A. V. Kiselev and V. I. Lygin, Infrared Spectra of Surface Compounds . Keter 
Publishing House Jerusalem Ltd., John Wiley & Sons, Inc., New York, 1975. 

86. L R. Snyder and J. W. Ward, The Surface Structure of Porous Silicas, 
Journal of Physical Chemistry, Vol. 70, 1966, p. 3941-3952. 

87. G. J. Young, Interaction of Water Vapor with Silica Surface, Journal of Colloid 
Science, Vol. 13, 1958, p.67-85. 

88. H. A. Benesi and A. C. Jones, An Infrared Study of the Water-Silica Gel System, 
Journal of Physical Chemistry, Vol. 63, 1957, p.179-182. 



150 



89. J. A. Hockey and B. A. Pethica, Surface Hydration of Silicas, Transactions of 
Faraday Society, Vol. 57, 1961, p. 2247-2262. 

90. A. V. Kiselev, Structure and Properties of Porous Materials, Colston Papers, Vol. 
10, Butterworth, London, 1958, p. 195. 

91 . R. S. McDonald, Surface Functionality of Amorphous Silica by Infrared 
Spectroscopy, Journal of Physical Chemistry, Vol. 62, 1958, p. 1168-1178. 

92. J. H. Anderson Jr. and K. A Wickersheim, Near Infrared Characterization of 
Water and Hydroxyl Groups on Silica Surfaces, Surface Science, Vol. 2, 1964, p. 

252-259. 

93. J. B. Peri, Infrared Study of OH and NH2 Groups on the Surface of a Dry Silica 
Aerogel, Journal of Physical Chemistry, Vol. 70, 1966, p. 2937-2945. 

94. N. W. Cant and L. H. Little, The Infrared Spectrum of Ammonia Adsorbed on 
Cab-O-Sil Silica Powder, Canadian Journal of Chemistry, Vol. 43, 1965, p. 
1252-1254. 

95. N. W. Cant and L H. Little, An Infrared Study of the Absorption of Ammonia on 
Porous Vycor Glass, Canadian Journal of Chemistry, Vol. 42, 1964, p. 802- 
809. 

96. M. L Hair and I. D. Chapman, Surface Composition of Porous Glass, Journal of 
the American Ceramic Society, Vol. 49, 1966, p. 651-654. 

97. T. H. Elmer, I. D. Chapman and M. E. Nordberg, Changes in Length and Infrared 
Transmittance During Thermal Dehydration of Porous Glass at Temperatures Up 
to 1200°C, Journal of Physical Chemistry, Vol. 66, 1962, p. 1517-1519. 

98. M. R. Basila, Hydrogen Bonding Interaction between Absorbate Molecules and 
Surface Hydroxyl Groups on Silica, Journal of chemical physics, Vol. 35, 1961, 
p. 1151-1158. 

99. V. Y. Davydov, L T. Zhuravlev and A. V. Kiselev, Study of the Surface and Bulk 
Hydroxyl Groups of Silica by Infrared Spectra and D20-Exchange, Transactions 
of Faraday Society, Vol. 60, 1964, p. 2254-2264. 

100. Jurgen R. Meyer-Arendt, Introduction tn Classical and Mode rn Optics. Prentice- 
Hall, Inc., New Jersey, 1984. 

1 01 . Ivan Fanderlik, Optical Properties of Gla ss. Glass Science and Technology 5. 
Elsevier, New York, 1983. 

102. Kurt Nassau, Part II: Color Involving Vibrations and Simple Excitations, in The 
Phvsics and Chemistry of Color-The Fifteen Causes of Color . Wiley-lnterscience 
Publication John Wiley & Sons, Inc., New York, 1983, p. 65-76. 

103 G. William Tasker and William G. French, Low-Loss Optical Waveguides with 
Pure Fused Si0 2 Cores, IEEE, Vol. 62, 1974, p. 1281-1282. 



251 



104. K. Susa, S. Satoh, I. Matsuyama, and T. Suganuma, New Optical Fiber Fabrication 
Method, Electron. Lett., Vol.18, No. 12, 1982, p. 499-500. 

105. Table of Periodic Properties of the Elements, Sargent-Welch Scientific Co., 
Skokie, Illinois. 

106. Kenzo Susa, Iwao Matsuyama, Shin Satoh and Tsuneo Suganuma, Reduction of 
Chlorine Content in Sol-Gel Derived Silica Glass, Journal of Non-Crystalline 
Solids, Vol.79, 1986, p. 165-176. 

107. Handbook of Chemistry and Physics, 64th ed., CRC Press, Inc., Boca Raton, 
Florida, 1 983-1 984, p. B-135. 

108. Neutron Activation Analysis of Trace Elements, Department of Nuclear 
Engineering North Carolina State University, Raleigh, North Carolina. 

109. C.S. Vikram, D. K. Agrawal, R. Roy and H. A. Mckinstry, A Simple Laser Speckle 
Dilatometer for Thermal Expansion Measurements, Material Letters, Vol. 3, 
No. 12, 1985, p. 482-484. 

1 1 0. ASTM C-730, in Annual Book of ASTM Standards . 1 976. 

111. Naval Publications and Forms Center, Departments and Agencies of the 
Department of Defense, Military Specification, Glass, Optical, MIL-G-174, 
Amendment 2, Philadelphia, 25 June 1974. 

112. A. Paul, Coloured Glasses, in Chemistry of Glasses . Chapman and Hall Ltd., 
New York, 1982, p. 204-270. 

113. S. Hufner, Chapter 1 , in Optical Spectra of Transparent Rare E arth Compounds. 
Academic Press, New York, 1978, p. 1-13. 

114. L E. Orgel, Introduction to Transition Metal Che mistry Liaand Field Theory. John 
Wiley & Sons, Inc., New York, 1960. 

115. W. A. Weyl, Coloured Glasses . Society of Glass Technology, Sheffield, England 
1951. 

116. T. Bates, Ligand Field Theory and Absorption Spectra of Transition-Metal Ions in 
Glasses, in Modern Aspects of the Vitreous State. Vol. 2 . J. D. Mackenzie ed., 
Butterworth, Inc., Washington DC, 1962, p. 195-254. 

1 1 7. Foster L. Harding, The Develo pmen t of Colors in Glass . Brackway Giass Co., Inc., 
Brockway, Pennsylvania. 

118. F. A. Cotton and C. Wilkinson, Advanced Inorganic Chemistry . 4th ed., John Wiley 
& Sons, Inc., New York, 1980. 

119. P. C. Schultz, Optical Absorption of the Transition Elements in Vitreous Silica, 
Journal of the American Ceramic Society, Vol. 57, July 1974, p. 309-313. 

120. O. G. Holmes and D. S. McClure, Optical Spectra of Hydrated Ions of the Transition 
Metals, Journal of Chemical Physics, Vol. 26, 1957, p. 1686-1694. 



BIOGRAPHICAL SKETCH 

Shi-Ho Wang received a B. S. in mineral and petroleum engineering from the 
National Cheng Kung University, Tainan, Taiwan, in 1976. Upon graduation, he was 
required to serve the nation two years by law as a politics and science instructor 
lieutenant in Guantiam Soldier Training Center, Army, Tainan, Taiwan. After serving in 
the Army, he was first employed as engineer and promoted to vice manager of the 
Engineering Department at Jong Mei Mineral Prospecting & Foundation Co., Taipei, 
Taiwan, in 1978. His duties involved quantitative analysis of mineral components and 
sampling design and engineering. Then he accepted a position as assistant engineer at the 
Department of Mines, Ministry of Economic Affairs, Taipei, Taiwan, in 1980 where his 
duties involved resolving the conflicts between domestic coke manufacturers and 
Japanese coke import agents, as well as issuing mining rights. 

For seeking a higher education, he was admitted to the Materials Science and 
Engineering Department of University of Florida as a graduate student in the spring 
semester of 1982. With his advisor Dr. Hench's encouragement, he passed the doctoral 
qualifying examination in the Fall semester, 1985. One year later he was a consultant 
and later chief research and development scientist at GelTech Inc., Alachua, Florida. 
Since September 1 987 he has devoted full time as a graduate associate to completing his 
doctoral degree in materials science and engineering. 



252 



I certify that I have read this study and that in my opinion it conforms to 
acceptable standards of scholarly presentation and is fully adequate, in scope and 
quality, as a dissertation for the degree of Doctor of Philosophy. 





Larry L. Hencty 

Graduate Research Professor of 

Materials Science and Engineering 

I certify that I have read this study and that in my opinion it conforms to 
acceptable standards of scholarly presentation and is fully adequate, in scope and 
quality, as a dissertation for the degree of Doctor of Philosophy. 



Jojephr H. Simmons 



Jd 

Professor of Materials Science 

and Engineering 

I certify that I have read this study and that in my opinion it conforms to 
acceptable standards of scholarly presentation and is fully adequate, in scope and 
quality, as a dissertation for the degree of Doctor of Philosophy. 



^^J^7~~ 




Vellayan Ramaswamy 

Professor of Electrical Engineering 

I certify that I have read this study and that in my opinion it conforms to 
acceptable standards of scholarly presentation and is fully adequate, in scope and 
quality, as a dissertation for the degree of Doctor of Philosophy. 




Gholamreza J. Abbaschian 
Professor of Materials 
Science and Engineering 



I certify that I have read this study and that in my opinion it conforms to 
acceptable standards of scholarly presentation and is fully adequate, in scope and 
quality, as a dissertation for the degree of Doctor of Philosophy. 



T^JLe (SLL- 



David E. Clark 
Professor of Materials 
Science and Engineering 



This dissertation was submitted to the Graduate Faculty of the College of Engineering 
and to the Graduate School and was accepted as partial fulfillment of the requirements 
for the degree of Doctor of Philosophy. 



April 1988 




Dean, Graduate School 






-